Risk Prediction Models for Falls Among Older Adults Inpatients with Cognitive frailty: Machine Learning Study Based on Comprehensive Geriatric Assessment

Risk Prediction Models for Falls Among Older Adults Inpatients with Cognitive frailty: Machine Learning Study Based on Comprehensive Geriatric Assessment | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Risk Prediction Models for Falls Among Older Adults Inpatients with Cognitive frailty: Machine Learning Study Based on Comprehensive Geriatric Assessment Lihua Chen, Meiwei Zhang, Weihua Yu, Xintong Liu, Yang Lü This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8022235/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background Falls are a major cause of disability in older adults, and cognitive frailty confers greater risk than isolated deficits. However, prediction models seldom target this subgroup. This study aimed to develop machine learning (ML)-based fall risk models for cognitively frail older adults using using Comprehensive Geriatric Assessment (CGA) data. Methods We included 814 hospitalized older adults with cognitive frailty, and corrected class imbalance using random under-sampling (n = 332). Eleven machine learning (ML) algorithms were trained using two feature selection strategies (top 100%–10% vs. bottom 90%–10%). Feature importance was evaluated through recursive feature elimination (RFE) and model-based approaches, with clinically actionable thresholds also determined. Results Seven key predictors were consistently identified across sampling strategies: First-Ever Fall-Related Injury (FH1-Injury), ADL (Activities of Daily Living) score, Age, Waist Circumference, Hearing Deficit, Generalized Anxiety Disorder-7 (GAD-7) score, and the Mini-Mental State Examination (MMSE) score. Lower ADL (< 100) and lower MMSE (< 19) scores were associated with increased fall risk, reflecting functional and cognitive decline. Likewise, advanced age (≥ 79 years), higher GAD-7 (≥ 1) scores indicating anxiety symptoms, and greater waist circumference (≥ 90 cm) predicted elevated fall probability. Decision tree, AdaBoost, and gradient boosting achieved near-perfect discrimination (AUC ≈ 1.00), even when limited to the top 20% of features. Logistic regression yielded comparably high accuracy and AUC while maintaining interpretability, making it suitable for clinical deployment. Conclusions This study presents a robust and scalable ML framework that integrates multidimensional CGA data to predict falls in cognitively frail older adults. Our findings support the development of a tailored fall risk scale and inform multidimensional interventions to prevent falls in this vulnerable population. Machine learning Prediction model Cognitive Frailty Falls Older adults Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 INTRODUCTION The rapid aging of the global population has led to an increased prevalence of age-related health issues, particularly falls, which are among the most common and preventable causes of injury-related morbidity and mortality for old adults [ 1 , 2 ]. According to the World Health Organization, an estimated 684,000 people die each year from falls worldwide [ 3 ]. Studies have shown that individuals over the age of 60 account for the highest proportion of fatal falls. Specifically, the incidence of falls ranges from 28% and 35% among those aged 65, and from 32% to 42% among individuals over 70 years of age [ 4 ]. Falls can lead to significant physical injuries, including fractures—especially hip fractures [ 5 ], head traumas, and even paralysis. These injuries can result in impaired physical function, prolonged rehabilitation periods and a diminished quality of life. Moreover, the financial burden of falls in older adults is substantial, with costs estimated to range from approximately $ 2,044 to $ 25,955 [ 6 ]. This highlights the urgent need for the early identification of individuals at risk for falls, which is essential for reducing fall incidents and improving overall health outcomes. Although the high incidence and severe outcomes of falls have been extensively documented, it is crucial to focus on the vulnerabilities of specific populations who are more prone to such incidents. Cognitive frailty refers to a combined condition characterized by the simultaneous presence of both physical frailty and cognitive impairment. The concept of cognitive frailty was introduced by an international consensus group from the International Academy of Nutrition and Aging (IANA) in collaboration with the International Association of Gerontology and Geriatrics (IAGG) in 2013. This highlighted the need for targeted interventions in this population [ 7 ]. Individuals with cognitive frailty had a 45% higher risk of falling compared to those without cognitive impairment [ 8 ]. Research has demonstrated that patients with cognitive frailty face greater challenges in walking and maintaining balance due to factors such as reduced muscle strength, diminished physical ability, impaired cognitive function, poor environmental adaptability, and social isolation [ 9 ]. Therefore, timely identification and intervention regarding fall risks in this population are essential. Despite the availability of various clinical assessment tools, such as the Tinetti Balance Assessment and Berg Balance Scale, these tools have limited predictive power in cognitively frail populations. They often fail to account for the multifaceted nature of fall risk in these individuals, which includes not only physical but also cognitive, nutritional, and psychological factors. Additionally, cultural nuances and socio-environmental contexts prevalent in Asian societies may further influence fall risk profiles and the effectiveness of generalized assessment tools. Consequently, there is a compelling need for more accurate and personalized predictive models specifically tailored for Asian older adults with cognitive frailty to address the shortcomings of existing tools in this context. The adoption of machine-learning (ML) models in clinical environments has gained significant attention due to its ability to formulate robust risk models and enhance predictive performance [ 10 ]. ML has emerged as a compelling alternative to traditional multiple linear regression (MLR) in medical research due to its ability to capture non-linear relationships and intricate interactions among numerous predictors without the assumption of a normal data distribution [ 11 ]. As a result, ML can potentially outperform conventional MLR in disease prediction [ 12 ]. While various machine learning methodologies for fall risk prediction have been explored in previous studies, models specifically targeting older adults with cognitive frailty, particularly within Asian populations, remains relatively under-investigated. To address these gaps, this study aims to develop a series of machine learning models to accurately predict the risk of falls in Asian older adult inpatients with cognitive frailty, utilizing data from Comprehensive Geriatric Assessment (CGA). By integrating a diverse array of variables encompassing cognitive function, physical frailty, nutritional status, and psychological well-being, we seek to enhance the precision of fall risk prediction and to identify critical determinants contributing to falls in this specific demographic. The resulting models will provide clinicians with effective prospective decision support tools, facilitating the timely and accurate identification of high-risk individuals for targeted early interventions, thereby reducing fall incidence and improving patient outcomes. This research holds significant practical implications for advancing geriatric healthcare management in Asian regions and addressing the challenges posed by population aging. METHODS Data source Data collection The datasets used for constructing the ML model were derived from the CGA Database of hospitalized patients in the geriatric department at The First Affiliated Hospital of Chongqing Medical University, covering the period from March 2016 to June 2023. Inclusion criteria were: (1) Age ≥ 60 years; (2) Complete assessment data available; (3) For patients with multiple assessments, only data from the first assessment were used. The excluded criteria were: (1) patients with only physical frailty, only cognitive impairment, or dementia; (2) patients under the age of 60 years; (3) patients with essential information missed or over 20% of the information lost. All participants underwent in-person assessments by professional evaluators. The evaluator held a postgraduate degree in nursing, was qualified as a comprehensive geriatric assessor through training, and had over twenty years of experience. The research methodology adheres to the principles of the Declaration of Helsinki and its revisions, as approved by the ethics board of The First Affiliated Hospital of Chongqing Medical University. Informed consent was waived for this retrospective study, which follows STROBE guidelines. An overview of the study workflow is presented in Fig. 1 . Definition and Assessment of Cognitive Frailty The diagnosis of cognitive frailty is based on the diagnostic criteria proposed by the International Academy on Nutrition and Aging (IANA) and the International Association of Gerontology and Geriatrics (IAGG) in 2013: (i) the presence of both physical frailty and cognitive impairment, with a Clinical Dementia Rating (CDR) of 0.5; and (ii) the exclusion of Alzheimer’s disease or other types of dementia. In our study, cognitive frailty is characterized by the concurrent presence of at least one of Fried's criteria and mild cognitive impairment as defined by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) Assessment of Physical Frailty Physical frailty is defined using Fried’s criteria, which involve five components: (i) unintentional weight loss, characterized by an involuntary loss of more than 5 kg in the past year; (ii) exhaustion, assessed through two self-reported questions from the Center for Epidemiological Studies Depression Scale: “How many times in the past week have you struggled to get anything done?” and “How often do you feel unable to move forward?” A response of “often” or “most of the time” to either question indicates exhaustion; (iii) weak muscle strength, evaluated by handgrip strength using an electronic hand dynamometer (Zhongshan Camry Electronic Co. Ltd, Guangdong, China). Participants stand and grip the dynamometer with maximum force using their dominant hand, with two trials conducted to record the maximum value. Grip strength less than 28 kg for men or less than 18 kg for women indicates reduced strength; (iv) slowness, assessed via a 6-meter fast gait speed test, where a speed of less than 1.0 m/s signifies slowness; and (v) low physical activity, determined by the question “How do you usually engage in physical exercise?” Responses such as “no physical exercise” or “mostly sedentary” indicate low activity levels. Each of the aforementioned criteria contributes one point. Participants are categorized as frail (three or more points), pre-frail (one or two points), or robust (zero points). To enhance the recognition of cognitive frailty, individuals in pre-frail states are also included in our analysis. Assessment of Cognitive Function Cognitive function was assessed using the Mini-Mental State Examination (MMSE), which provides a comprehensive and accurate evaluation of a subject’s intellectual status and degree of cognitive impairment. The MMSE encompasses seven dimensions: orientation to time, orientation to place, immediate memory, delayed memory, attention and calculation, language, and visual-spatial abilities. The MMSE is a 30-item scale with scores ranging from 0 to 30, where a higher score indicates better cognitive function. Participants with intact cognition are defined as those with an MMSE score of 27 or greater and no clinical diagnosis of cognitive impairment. For patients with mild cognitive impairment (MCI), the diagnostic criteria are based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) by the American Psychiatric Association (2013): (i) subjective and objective examinations indicating MCI; (ii) cognitive decline in one or more of the aforementioned dimensions; (iii) preserved daily living abilities; (iv) absence of dementia diagnosis; (v) exclusion of other conditions causing cognitive decline; and (vi) MMSE scores. The MMSE scores are adjusted based on education level, with cut-off points of 17 for illiterate individuals, 20 for primary education, 22 for junior high education, and 23 for university-level or higher education. Outcome Measure The primary outcome measure was the occurrence of falls, defined by the variable “Fall History” (FH), which indicates whether a participant experienced a fall in the past two years. Additionally, the variable “First-Ever Fall-Related Injury” (FH1-Injury), a binary indicator (Yes/No), captured whether the participant’s first fall resulted in an injury requiring medical attention. This variable reflects the historical severity of the participant’s first lifetime injurious fall, which could have occurred long before the observation window for FH. Importantly, FH1-Injury represents the first-ever injurious fall, making it temporally independent of the study’s observation period for recent falls. Predictive Variables In this study, 93 candidate predictor variables were initially considered, including demographic factors (age, gender, lifestyle), clinical indicators (MMSE, PSQI), and functional characteristics (vision impairment, ADL scores). To address potential redundancy and multicollinearity, Principal Component Analysis (PCA) was applied, reducing dimensionality and creating composite features such as PCA_Hip_Circ_Waist_Circ and PCA_Hear_Act_Hear_Def. Pearson correlation analysis revealed significant correlations between variables (e.g., Waist_Circ and Hip_Circ: 0.731; Hear_Act and Hear_Def: 0.89). After preprocessing, 58 variables were retained for final analysis. A detailed list and descriptions of all predictor variables are provided in Supplementary Material 1. Statistical Analysis Continuous variables were shown as mean ± SD with range or median with IQR, based on distribution. Group comparisons used the Student’s t-test for normal data or Wilcoxon rank-sum test for non-normal data. Categorical variables were presented as frequencies and percentages, with comparisons using the chi-square test or Fisher’s exact test. Descriptive and variance analyses were conducted with SPSS Version 26.0, while data preprocessing, feature selection, ML model development, and evaluation were carried out using Python 3.8. Statistical significance was set at p < 0.05. Data processing The overall workflow of data preprocessing, feature selection, model construction, and performance evaluation is illustrated in Fig. 1 A. Specifically, data preprocessing involved addressing missing values, normalizing continuous variables, and encoding categorical variables. To balance the number of fall and non-fall cases, random under-sampling was applied. Six feature selection methods—Pearson Correlation, Chi-square test, Recursive Feature Elimination (RFE), Logistic Regression, Random Forest, and Light Gradient Boosting Machine (Light GBM)—were utilized to identify the most relevant predictors of fall risk. Feature selection is performed on six feature sequences, F1, F2, F3, F4, F5, and F6 (F1-F6 refer to the ordered sequence of features in the different selection methods, more detailed steps were seen in the Supplementary Material 2, section A), followed by model construction. The final feature sequence F: $$\:F=\sum\:_{i=1}^{i=6}{W}_{i}{F}_{i}$$ Here, \(\:{W}_{i}\) represents the weight parameter corresponding to the i-th feature analysis method, taking a value of 0 or 1, where 1 indicates retention of the feature, and 0 denotes its exclusion. After feature selection, 58 features were retained, and the features were subsequently grouped based on their importance rankings for further analysis. Algorithms A total of eleven ML algorithms were employed to build predictive models for each feature subset. The algorithms used included Decision Tree, Gradient Boosting, Quadratic Discriminant Analysis (QDA), AdaBoost, Naive Bayes, Logistic Regression, Random Forest, Linear Discriminant Analysis (LDA), Neural Network, and Support Vector Machines (SVM), as well as k-Nearest Neighbors (k-NN). Each machine learning algorithm was applied to every feature subset, and the results from all datasets were compiled for comparison and analysis. Detailed procedures for model development are presented in Supplementary Material 2, Section B, and the parameter tuning strategies for the machine learning algorithms are described in Supplementary Material 2, Section C table. Model evaluation indexes The primary metric for model performance evaluation was the area under the receiver operating characteristic curve (AUC-ROC). In addition, various performance metrics including accuracy, sensitivity, specificity, recall, precision, and F1 score were also examined. All statistical analyses and modeling procedures were conducted using Python (version 3.8), with statistical significance set at a two-sided P-value of less than 0.05. To avoid over-fitting and to provide an unbiased estimate of generalisation performance, we adopted a nested, repeated 10-fold cross-validation (CV). A two-stage procedure was used: (i) outer loop (model assessment) – a stratified 10-fold CV repeated three times (30 train/test splits in total) delivered point estimates and 95% CIs of each metric; (ii) inner loop (model selection) – for every training partition in the outer loop, a further stratified 10-fold CV performed random/grid search over the hyper-parameter space. Feature imputation, scaling, and the feature-selection step were treated as components of a single Pipeline object and re-computed inside the inner loop to prevent information leakage. The tuning objective was ROC-AUC; ties were broken by favouring the more parsimonious model (smaller tree depth, fewer estimators, or lower C). All experiments were implemented with scikit-learn 1.5.0, imbalanced-learn 0.12.0, and NumPy 1.26; random seed fixed at 42 for full reproducibility. A detailed description of model performance evaluation is presented in Supplementary Material 2, Section D. Results 1. Demographic and Clinical Characteristics A total of 814 participants were included in the study based on the inclusion criteria. To address the significant sample size imbalance between the fallers and non-fallers, we applied a random under-sampling technique, resulting in the exclusion of 432 participants (Figure 1B). The initial dataset is referred to as the "full dataset" (n = 814), while the dataset after under-sampling is referred to as the "Under-sampled Balanced Dataset " (n = 332). 1.1 Full Dataset (n = 814) As shown in Table 1A, the demographic and clinical characteristics of fallers and non-fallers in the full dataset demonstrated significant group differences. Fallers (n = 166) were significantly older than non-fallers (n = 648), with a mean age of 78.05 ± 7.31 years versus 76.32 ± 7.92 years, respectively ( P = 0.006). Other demographic characteristics, such as gender and educational level, did not show significant differences between the two groups ( P > 0.05). Notably, fallers had higher rates of osteoporosis (31.3% vs. 21.1%, P = 0.013) and visual impairment (88.6% vs. 79.0%, P = 0.005) compared to non-fallers. Additionally, fallers were more likely to have vision problems affecting daily activities (24.1% vs. 12.96%, P < 0.0001) and urinary incontinence (24.1% vs. 15.4%, P = 0.008). 1.2 Under-sampled Balanced Dataset (n = 332) Correspondingly, Table 1B presents the demographic characteristics in the balanced dataset, showing a similar pattern of group differences after random under-sampling. Fallers were also significantly older ( P = 0.028) and had higher rates of osteoporosis (31.3% vs. 16.9%, P = 0.008) and visual impairment (88.6% vs. 80.1%, P = 0.035). Urinary incontinence remained a distinguishing factor between fallers and non-fallers (24.1% vs. 15.1%, P = 0.038). However, differences in smoking status and gender distribution were less pronounced in this dataset compared to the full sample. 2. Key Predictors of Fall Risk Across both the fully sampled dataset (n = 814) and the under-sampled balanced dataset (n = 332), several features consistently emerged as significant predictors of fall risk. These predictors were identified through feature selection methods (Pearson/χ², RFE) and model-based importance ranking (logistic regression, random forest, LightGBM). The top-ranked predictors spanned four key domains: vision function (Vision_Act, Vision_Def), sleep characteristics (Sleep_Time, Sleep_Duration, PSQI), functional and cognitive status (lower ADL, slightly lower MMSE), and frailty/multimorbidity (Fried stage, ≥2 chronic conditions), with urinary incontinence and BMI. Across both datasets, the importance rankings of most features were largely consistent. Table 2 summarizes the scores of the top 15 features identified in both the full and balanced datasets. Dataset-specific differences were minimal: Age and residence area were slightly more relevant in the balanced dataset but did not affect the overall ranking or the direction of associations. Collectively, these findings across different methods and datasets indicate a stable, multifactorial risk profile for falls in cognitively frail older adults. The complete ranking results for all 58 features in both datasets are provided in Supplementary Material 3 (Section A for full dataset, section B for balanced dataset). 3. Model Development and Performance Across Feature Subsets We developed fall risk prediction models using 11 machine learning algorithms, evaluated across different feature proportions, ranging from the full feature set (100%) to the top 10%, as well as the bottom 90% to bottom 10% subsets. Figures 2 and 3 present the ROC curves for all machine learning classifiers under different feature subsets in the full and balanced datasets, respectively. In the full dataset (Figure 2), models trained on the top-ranked features (Panel A) consistently achieved high discrimination, with ensemble methods (AdaBoost, Gradient Boosting, Random Forest, Decision Tree) and Logistic Regression yielding AUCs close to 1.00 when ≥20% of features were included. In contrast, models trained on the bottom-ranked features (Panel B) showed a rapid decline in discriminative ability, with ROC curves approaching the diagonal line as feature subsets decreased, underscoring the limited predictive value of low-ranked variables. Similarly, in the balanced dataset (Figure 3), the top feature subsets (Panel A) sustained strong predictive performance across classifiers, with AUCs consistently >0.90 when at least 20% of features were retained. However, bottom feature subsets (Panel B) again demonstrated markedly reduced performance, with AUCs clustering around 0.5, highlighting the robustness of the top-ranked predictors even under class-balancing conditions. To further quantify these findings, Figure 4 complements the ROC curves by providing a comparative overview of AUC-ROC values (Panels a, c) and F1-scores (Panels b, d) across top and bottom feature subsets. The line plots clearly show that AUC remained near-perfect when using the top 20% of features, whereas performance declined sharply with bottom-ranked subsets. The corresponding bar plots reinforce this observation, demonstrating that models achieved the highest F1-scores with top features, while bottom subsets led to unstable and substantially lower F1 values. Finally, these graphical patterns are substantiated in Table 3, which summarizes the performance of all 11 algorithms under the full, 20%, and 10% top feature subsets. The performance of all models across every feature proportion (from Top100% to Top10% and Bottom90% to Bottom10%) is provided in Supplementary Material 4. Ensemble methods (AdaBoost, Gradient Boosting, Random Forest) and Decision Trees consistently achieved near-perfect AUC (≈1.00) with the top 100% and 20% features, while Logistic Regression offered comparable accuracy with greater interpretability. However, when restricted to only the top 10% of features, performance variability across algorithms increased, and F1-scores were generally reduced. This integrated analysis across ROC curves, line/bar plots, and tabular results provides convergent evidence that top-ranked features not only preserve predictive accuracy but also yield a more stable and clinically interpretable fall-risk model. 4. Feature Importance and Clinical Thresholds Figure 5 illustrates the feature importance rankings derived from random forest models after training. Across both the fully sampled (figure 5a) and under-sampled balanced datasets (figure 5b), seven variables consistently emerged as the most influential predictors of fall risk: history of first injurious fall (FH1-Injury), ADL score, age, waist circumference, hearing deficit, GAD-7 score, and MMSE score. The consistent ranking across sampling strategies indicates the stability of these predictors. Additional contributors, such as vision impairment, sleep quality indicators (e.g., PSQI score, sleep duration, latency), and frailty status (e.g., Fried stage), further highlight the multifactorial nature of fall risk in cognitively frail older adults. To enhance robustness, we performed supplementary analyses using random forest–based impurity reduction and neural network–based permutation importance. Both approaches supported the stability of the identified predictors, although minor variability and occasional negative values were observed with permutation importance. Such fluctuations likely reflect methodological sensitivity to noise, overfitting, or multicollinearity. Importantly, the consistency of results across methods provides convergent evidence for the reliability of these core predictors. To improve interpretability, the direction and threshold effects of key predictors were examined. A lower ADL score (<100) and lower MMSE score (<19) were both associated with higher fall risk, reflecting reduced functional and cognitive capacity. Similarly, advanced age (≥79 years) and elevated GAD-7 scores (≥1), reflecting greater anxiety symptoms, were associated with a higher likelihood of falls. In contrast, better functional independence, preserved cognitive performance, and psychological stability were protective against fall risk. These threshold patterns remained consistent across analytical models, reinforcing the robustness and reproducibility of these key predictors. Discussion This study provides compelling evidence that machine learning (ML) applied to comprehensive geriatric assessment (CGA) data can yield highly accurate and clinically meaningful fall risk predictions in older adults with cognitive frailty within two years. By integrating both full-sampling and under-sampling strategies across multiple ML methods, we identified a robust and reproducible set of seven core predictors of fall risk: history of first injurious fall, activities of daily living (ADL) score, age, waist circumference, hearing deficit, GAD-7 score, and MMSE score. For several of these, we determined clinically actionable threshold values (e.g., ADL = 100, age = 79 years, MMSE = 19) beyond which fall risk markedly increases. These threshold-based insights provide a foundation for individualized risk stratification and targeted intervention in this high-risk population. Key Predictors and Clinical Relevance: Each of the core predictive features is strongly associated with known biological mechanisms and clinical manifestations of fall risk in cognitively frail older adults. A history of an injurious fall emerged as an especially influential predictor, aligning with findings by Chen et al[ 13 ],. that prior falls are a key risk factor for future falls in community-dwelling elders. An initial injurious fall can lead to functional decline, manifested as reduced muscle strength, impaired balance, and limited joint mobility. It often induces a fear of falling, which in turn reduces physical activity and further impairs gait stability. This vicious cycle increases the risk of subsequent falls and recurrent injuries. The ADL score was another consistently top-ranked predictor; lower ADL scores indicate greater functional dependency and frailty, reflecting declines in physical capacity and independence [ 14 , 15 ]. Psychological status, as measured by the GAD-7, was also a strong contributor to the model. Higher anxiety levels were associated with increased fall risk, which is biologically plausible given that anxiety and related affective disturbances can impair attention, psychomotor coordination, and executive function, all of which are essential for safe mobility. This observation is consistent with the report by Wang et al. [ 16 ] linking anxiety and depression to significantly higher risks of falls and fall-related injuries in older adults. Waist circumference showed a positive association with fall risk in our study. Notably, 7.3% of participants had a waist circumference exceeding 90 cm (mean ± SD: 86.7 ± 10.8 cm). Although the odds ratio at this threshold was modest and not statistically significant (OR = 1.03, 95% CI: 0.73–1.46), we considered a 90 cm waist as a potentially important cutoff for clinical attention. Central obesity, as indicated by an excessive waistline, is known to impair balance and restrict mobility [ 17 , 18 ]. Thus, a waist circumference above 90 cm may serve as a practical marker for elevated fall risk in cognitively frail older adults, though this proposition requires validation in larger cohorts. Hearing deficit also emerged as a key predictor in our models. Sensory impairments such as hearing loss can diminish situational awareness and increase cognitive load during ambulation, hampering an individual’s ability to maintain balance and detect environmental hazards. This mechanism is supported by epidemiological studies [ 19 , 20 ] and by Gopinath et al. [ 21 ], who reported significantly higher fall rates among older adults with multisensory impairments (e.g., concurrent hearing and vision loss). Cognitive deficits contributed significantly to fall risk as well. In addition to hearing loss, global cognitive function (assessed by the MMSE) was a consistent predictor in our analysis. We identified an optimal MMSE cutoff score of 19 (Youden’s index = 0.027; sensitivity 22.4%, specificity 80.3%), beyond which fall risk rose appreciably. Although the discriminatory power of this specific threshold was modest, it aligns with the notion that even moderate cognitive impairment can heighten fall susceptibility. Consistent with our findings, a previous study [ 22 ],found that cognitive decline in older adults was significantly associated with an increased incidence of serious fall-related injuries. These observations underscore the importance of incorporating routine cognitive assessments into fall risk evaluation frameworks for cognitively frail individuals, as cognitive deficits are a known contributor to falls. Taken together, our findings illustrate the multifactorial nature of fall risk in older adults with cognitive frailty, spanning physical, psychological, sensory, and cognitive domains. A prior injurious fall reflects both functional deterioration and psychological consequences (e.g., fear of falling), lower ADL scores signal loss of independence, and elevated anxiety levels highlight the role of affective symptoms in impairing balance and coordination. Likewise, sensory deficits (especially hearing loss) reduce environmental awareness, while central adiposity (large waist circumference) compromises stability. This comprehensive perspective reinforces that elevated fall risk arises from an interplay of diverse factors, underlining the need for multidimensional strategies in risk assessment and prevention. Model Performance and Methodological Advancements: Beyond identifying risk factors, our machine learning framework achieved outstanding predictive performance. Most classifiers demonstrated near-perfect discrimination (AUC ≈ 1.00) even when restricted to the top 20% of features. Notably, logistic regression achieved comparably high accuracy and AUC while providing greater transparency and interpretability, which is critical for clinical use. This performance clearly exceeds that of conventional fall risk models, which typically report AUCs of 0.65–0.88 (median ~ 0.72 [ 23 ],); for instance, a recently published model achieved an AUC-ROC of only 0.734 [ 24 ]. Our findings also extend those of Park et al. (2025) [ 25 ], who used logistic regression on longitudinal data from the Korean Frailty and Aging Cohort Study to identify four optimal variables (Fried PF phenotypes, PF-M, SGDS-K, and SARC-F), achieving excellent discrimination (AUC, sensitivity, specificity, and accuracy ≥ 91%). While their work demonstrated the feasibility of using a small set of physical and psychological variables, our study incorporated a broader pool of 93 CGA-derived candidate features, from which 58 were retained after rigorous feature selection. We further compared 11 machine learning algorithms under both full- and under-sampled conditions, ensuring that model stability was rigorously assessed across class distributions. Unlike most prior studies that relied on oversampling techniques (e.g., SMOTE), our use of random under-sampling reduced majority-class bias and improved generalizability. Several methodological innovations likely contributed to these excellent results, distinguishing our work from traditional models [ 26 , 27 ],. First, we integrated a broad range of multidimensional predictors, combining conventional clinical markers (e.g., known osteoporosis, ADL scores) with psychological, sensory, and lifestyle variables to capture a more holistic profile of fall risk. This inclusive approach goes beyond the narrower, expert-selected feature sets often used in earlier fall risk models. Second, we employed both incremental and decremental feature selection strategies to optimize model parsimony without sacrificing accuracy. Our results demonstrate that high discrimination can be maintained with a substantially reduced feature set, which is critical for real-world feasibility and efficiency in clinical settings. Third, we explicitly addressed class imbalance by applying random under-sampling of the majority class (non-fallers). This procedure improved model generalizability and mitigated bias toward the majority class, an issue often overlooked in previous fall risk studies where imbalanced outcomes can lead to overestimation of model performance. Together, these strategies enhanced our model’s robustness, efficiency, and clinical applicability. Strengths and Limitations: This study has several notable strengths. It is one of the first to focus specifically on fall risk prediction in cognitively frail older adults—a high-risk subgroup that is often underrepresented in fall prevention research. By leveraging comprehensive CGA data, we were able to integrate a wide array of features across medical, functional, psychological, and sensory domains, which strengthens the ecological validity of our findings. Additionally, we rigorously evaluated multiple modeling techniques and sampling strategies (full-sample vs. under-sampling), which adds confidence in the consistency and reproducibility of the identified predictors and model performance. The use of both complex ensemble models and simpler interpretable models (like logistic regression) highlights the versatility and scalability of our approach for different clinical scenarios. Nevertheless, several limitations must be acknowledged. First, our study was conducted in a single-center geriatric inpatient cohort, which may limit the external generalizability of the findings. The participant population was relatively homogeneous, and model performance should be validated in community-dwelling older adults and more diverse geographic or ethnic populations. Second, the study design was cross-sectional, meaning predictors and fall outcomes were assessed contemporaneously. This limits causal inference—longitudinal studies are needed to establish temporal relationships and confirm that the identified risk factors indeed precede and predict falls over time. Third, although we observed near-perfect classification performance, the modest sample size and the use of under-sampling techniques raise the possibility of overfitting. Our models may have captured patterns specific to this dataset that do not generalize universally. Therefore, external validation in larger, independent cohorts is essential to verify the stability of the predictor set and the true predictive accuracy of the models. Future research should also explore prospective validation and calibration of these models, as well as the integration of additional relevant features (e.g., gait or balance assessments) that were beyond the scope of our current dataset. Conclusion Overall, this study highlights that machine learning applied to CGA data can provide robust, scalable, and clinically relevant tools for fall risk prediction in cognitively frail older adults. By identifying a reproducible set of core predictors with actionable thresholds, our findings support the development of a specialized fall risk scale and supports the implementation of targeted, multidimensional interventions aimed at reducing falls in this high-risk population. Declarations Contributors Lihua Chen and Meiwei Zhang were responsible for developing the models and drafting the manuscript. Xintong Liu undertook data collection. Yang Lü and Weihua Yu contributed expertise in clinical study design and revise the manuscript. All authors read and approved the final manuscript. Data sharing statement The dataset, code, algorithm files, and de-identified results used in this study are not publicly available. However, the data for this study can be shared upon reasonable request to the corresponding author. Declaration of interests The authors declare that they have no potential competing interests. Ethics approval and consent to participate The study was approved by the Ethics Committee of The First Affiliated Hospital of Chongqing Medical University (approved on 18 July 2012, No.15) and has been performed in accordance with the ethical standards laid down in the Declaration of Helsinki and its later amendments. The need for written informed consent to participate was waived by the First Affiliated Hospital of Chongqing Medical University ethics committee due to retrospective nature of the study (institutional ethics board approved No: 2012-15) Clinical Trial Number Clinical trial number: not applicable. FUNDING This study was supported by grants from Chongqing Talent Plan (cstc2022ycjh-bgzxm0184), Key Project of Technological Innovation and Application Development of Chongqing Science & Technology Bureau (CSTC2021jscx-gksb-N0020), Science Innovation Programs Led by the Academicians in Chongqing under Project (2022YSZX-JSX0002CSTB), Chongqing Medical Key Discipline and Regional Medical Key Discipline Development Project 0201【2022】No. 144 202325 and Program for Youth Innovation in Future Medicine, Chongqing Medical University (W0166). References Older Adult Fall Prevention. CDC’s injury center uses data and research to save lives. Centers for Disease Control and Prevention website. [https://www.cdc.gov/falls/pdf/CDC-DIP_At-a-Glance_Falls_508.pdf] Colón-Emeric CS, McDermott CL, Lee DS, Berry SD: Risk Assessment and Prevention of Falls in Older Community-Dwelling Adults: A Review . Jama 2024, 331 (16):1397-1406. Burzyńska M, Kopiec T, Pikala M: Mortality Trends due to Falls in the Group of People in Early (65–74 Years) and Late (75+) Old Age in Poland in the Years 2000–2020 . International journal of environmental research and public health 2023, 20 (6):5073. Nicholson WK, Silverstein M, Wong JB, Barry MJ, Chelmow D, Coker TR, Davis EM, Jaén CR, Krousel-Wood M, Lee S et al : Interventions to Prevent Falls in Community-Dwelling Older Adults: US Preventive Services Task Force Recommendation Statement . Jama 2024, 332 (1):51-57. Kim KM, Lui L-Y, Cummings SR: Recent fall and high imminent risk of fracture in older men and women . Age and ageing 2022, 51 (6):afac141. Wen P, Sun L, Cheng L, Zhou C, Wang Z, Wang Z: Fall status and risk factors in older Chinese adults: A cross ‐sectional study . Journal of clinical nursing 2024, 33 (7):2609-2618. Kelaiditi E, Cesari M, Canevelli M, Van Kan GA, Ousset P-J, Gillette-Guyonnet S, Ritz P, Duveau F, Soto M, Provencher V: Cognitive frailty: rational and definition from an (IANA/IAGG) international consensus group . The Journal of nutrition, health and aging 2013, 17 (9):726-734. Wang S, Wang M, Leung ISH, Ge S, Xu X, Chen S, Yin Y: The association of cognitive frailty and the risk of falls among older adults: A systematic review and meta-analysis . Int J Nurs Pract 2023, 29 (4):e13181. Demanze Laurence B, Michel L: The fall in older adults: physical and cognitive problems . Current aging science 2017, 10 (3):185-200. Haug CJ, Drazen JM: Artificial intelligence and machine learning in clinical medicine, 2023 . New England Journal of Medicine 2023, 388 (13):1201-1208. Liu C-H, Peng C-H, Huang L-Y, Chen F-Y, Kuo C-H, Wu C-Z, Cheng Y-F: Comparison of multiple linear regression and machine learning methods in predicting cognitive function in older Chinese type 2 diabetes patients . BMC neurology 2024, 24 (1):11. Liu C-H, Peng C-H, Huang L-Y, Chen F-Y, Kuo C-H, Wu C-Z, Cheng Y-F: The comparison between multiple linear regression and machine learning methods in predicting cognitive function in Chinese type 2 diabetes . 2023. Xiaodong C, Lingxiao H, Kewei S, Yafei W, Shaowu L, Ya F: Interpretable Machine Learning for Fall Prediction Among Older Adults in China . Am J Prev Med 2023, 65 (4). Andrews J, Gold L, Reed M, Garcia J, McClelland R, Fitzpatrick A, Hough C, Cawthon P, Covinsky K: Appendicular Lean Mass, Grip Strength, and the Development of Hospital-Associated Activities of Daily Living Disability Among Older Adults in the Health ABC Study . The journals of gerontology Series A, Biological sciences and medical sciences 2022, 77 (7):1398-1404. Qin K, Lin L, Lu C, Chen W, Guo V: Association between systemic inflammation and activities of daily living disability among Chinese elderly individuals: the mediating role of handgrip strength . Aging clinical and experimental research 2022, 34 (4):767-774. Wang J, Li S, Hu Y, Ren L, Yang R, Jiang Y, Yu M, Liu Z, Wu Y, Dong Z et al : The moderating role of psychological resilience in the relationship between falls, anxiety and depressive symptoms . J Affect Disord 2023, 341 :211-218. Madigan M, Rosenblatt NJ, Grabiner MD: Obesity as a factor contributing to falls by older adults . Current obesity reports 2014, 3 :348-354. Matter KC, Sinclair SA, Hostetler SG, Xiang H: A comparison of the characteristics of injuries between obese and non-obese inpatients . Obesity (Silver Spring) 2007, 15 (10):2384-2390. Gopinath B, Rochtchina E, Wang JJ, Schneider J, Leeder SR, Mitchell P: Prevalence of age-related hearing loss in older adults: Blue Mountains Study . Arch Intern Med 2009, 169 (4):415-416. Skalska A, Wizner B, Piotrowicz K, Klich-Rączka A, Klimek E, Mossakowska M, Rowiński R, Kozak-Szkopek E, Jóźwiak A, Gąsowski J et al : The prevalence of falls and their relation to visual and hearing impairments among a nation-wide cohort of older Poles . Exp Gerontol 2013, 48 (2):140-146. Gopinath B, McMahon CM, Burlutsky G, Mitchell P: Hearing and vision impairment and the 5-year incidence of falls in older adults . Age and Ageing 2016, 45 (3):409-414. Muir SW, Gopaul K, Montero Odasso MM: The role of cognitive impairment in fall risk among older adults: a systematic review and meta-analysis . Age and ageing 2012, 41 (3):299-308. Dormosh N, van de Loo B, Heymans MW, Schut MC, Medlock S, van Schoor NM, van der Velde N, Abu-Hanna A: A systematic review of fall prediction models for community-dwelling older adults: comparison between models based on research cohorts and models based on routinely collected data . Age and Ageing 2024, 53 (7). Chen X, He L, Shi K, Wu Y, Lin S, Fang Y: Interpretable Machine Learning for Fall Prediction Among Older Adults in China . Am J Prev Med 2023, 65 (4):579-586. Park C, Kim N, Kim M, Won CW, Lee BC: Advancing fall risk prediction in older adults with cognitive frailty: A machine learning approach using 2-year clinical data . PLoS One 2025, 20 (8):e0330672. Van De Loo B, Seppala LJ, Van Der Velde N, Medlock S, Denkinger M, De Groot LC, Kenny R-A, Moriarty F, Rothenbacher D, Stricker B: Development of the AD F ICE_IT models for predicting falls and recurrent falls in community-dwelling older adults: pooled analyses of European cohorts with special attention to medication . The Journals of Gerontology: Series A 2022, 77 (7):1446-1454. Kang L, Chen X, Han P, Ma Y, Jia L, Fu L, Yu H, Wang L, Hou L, Yu X et al : A Screening Tool Using Five Risk Factors Was Developed for Fall-Risk Prediction in Chinese Community-Dwelling Elderly Individuals . Rejuvenation Res 2018, 21 (5):416-422. Tables Table 1. Comparison of Demographic and Clinical Characteristics Between Fallers and Non-Fallers in Full and Balanced Datasets Table 1A. Comparison of demographic, lifestyle, and health-related characteristics between fallers and non-fallers in the full dataset (n = 814) Variables Fully Sampled Dataset (n=814) P-value Non-fallers (n=648) Fallers (n=166) Demographic Characteristics Age (years), mean ± SD 76.32 ± 7.92 78.05 ± 7.31 0.006 Gender (Male), n (%) 266 (41.05%) 61 (36.75%) 0.313 Education level, n (%) 0.699 Illiteracy 118 (18.21%) 29 (17.47%) Primart school 157 (24.23%) 40 (24.10%) Junior school 288 (44.44%) 65 (39.16%) University or above 85 (13.12%) 32 (19.28%) Lifestyle and Health Behavior Smoking status, n (%) 0.028 Never smoked 488 (75.31%) 141 (84.94%) Previous smoker 87 (13.43%) 15 (9.04%) Current smoker 73 (11.27%) 10 (6.02%) Drinking status, n (%) 0.352 Never drank 542 (83.64%) 146 (87.95%) Previous drinker 56 (8.64%) 10 (6.02%) Current drinking 50 (7.72%) 10 (6.02%) Lifestyle, n (%) 0.676 Basic regular 473 (72.99%) 114 (68.67%) Highly regular 116 (17.90%) 34 (20.48%) Irregular 59 (9.10%) 18 (10.84%) Chronic Diseases and Health Conditions Coronary heart disease, n (%) 226 (34.88%) 58 (34.94%) 0.922 Hypertension, n (%) 393 (60.65%) 104 (62.65%) 0.758 Osteoporosis, n (%) 137 (21.14%) 52 (31.33%) 0.013 Stroke, n (%) 140 (21.60%) 47 (28.31%) 0.166 Diabetes, n (%) 227 (35.03%) 59 (35.54%) 0.619 Cataract, n (%) 53 (8.18%) 26 (15.66%) 0.014 Visual Impairment, n (%) 512 (79.01%) 147 (88.55%) 0.005 Vision Problems Affecting Daily Activities, n (%) 84 (12.96%) 40 (24.10%) 0.000001 Hearing Impairment, n (%) 290 (44.75%) 73 (43.98%) 0.857 Nutritional and Physical Status MNA Screening Result, n (%) 0.429 Well-nourished 418 (64.51%) 116 (69.88%) At risk of malnutrition 212 (32.72%) 46 (27.71%) Malnutrition 18 (2.78%) 4 (2.41%) Gastritis or Ulcer, n (%) 131 (20.22%) 21 (12.65%) 0.049 Waist Circumference(cm), mean ± SD 86.68 ± 10.44 86.81 ± 11.98 0.686 BMI (kg/m²), n (%) 0.380 < 18.5 54 (8.33%) 18 (10.84%) 18.5–24.9 368 (56.79%) 84 (50.60%) 25–29.9 212 (32.72%) 60 (36.14%) ≥ 30 14 (2.16%) 4 (2.41%) Sleep and Mental Health PSQI Score, mean ± SD 10.62 ± 5.29 * 9.46 ± 5.38 0.015 Sleep Duration (hours), n (%) 0.002 7 160 (24.69%) 51 (30.72%) Sleep Latency (minutes), n (%) 0.010 ≤ 15 220 (33.95%) 77 (46.39%) 16~30 131 (20.22%) 24 (14.46%) 31~60 87 (13.43%) 25 (15.06%) ≥ 60 210 (32.41%) 40 (24.10%) GDS-5 (depression), n (%) 201 (31.02%) 43 (25.90%) 0.199 GAD-7 (anxiety), n (%) 221 (34.10%) 43 (25.90%) MMSE Score, mean ± SD 22.52 ± 4.85 21.83 ± 5.07 0.094 ADL Score, mean ± SD 96.27 ± 4.50 94.55 ± 5.16 0.000066 Multi-Morbidity and Frailty Multiple Chronic Conditions (≥2), n (%) 578 (89.20%) 154 (92.77%) 0.172 Frailty Status, n (%) 0.001 Pre-frailty, n (%) 487 (75.15%) 103 (62.05%) Frailty, n (%) 161 (24.85%) 63 (37.95%) Limbs Flutter, n (%) 69 (10.65%) 26 (15.66%) 0.073 Urinary Incontinence, n (%) 100 (15.43%) 40 (24.10%) 0.008 Table 1B. Comparison of demographic, lifestyle, and health-related characteristics between fallers and non-fallers in the balanced dataset (n = 332) Variables Undersampled Data (n=332) Nonfallers (n=166） Fallers (n=166） P-value Demographic Characteristics Age (years), mean ± SD 76.30 ± 8.17 78.05 ± 7.31 0.028 Gender (Male), n (%) 76 (45.78%) 61 (36.75%) 0.094 Education level, n (%) 0.084 Illiteracy 29 (17.47%) 29 (17.47%) Primart school 33 (19.88%) 40 (24.10%) Junior school 85 (51.20%) 65 (39.16%) University or above 19 (11.45%) 32 (19.28%) Residence area, n (%) 0.091 Urban 113 (68.07%) 126 (75.90%) Suburban 6 (3.61%) 7 (4.22%) County town 31 (18.67%) 15 (9.04%) Village 16 (9.64%) 18 (10.84%) Living arrangement, n (%) 0.549 With spouse or children 150 (90.36%) 143 (86.14%) Living alone 11 (6.63%) 15 (9.04%) Nursing home 5 (3.01%) 7 (4.22%) Job category, n (%) 0.851 Primarily physical labor 88 (53.01%) 86 (51.81%) Primarily mental labor 66 (39.76%) 70 (42.17%) Others 12 (7.23%) 10 (6.02%) Lifestyle and Health Behavior Smoking status, n (%) 0.001 Never smoked 114 (68.67%) 142 (85.54%) Previous smoker 27 (16.27%) 14 (8.43%) Current smoker 25 (15.06%) 10 (6.02%) Drinking status, n (%) 0.194 Never drank 139 (83.73%) 147 (88.55%) Previous drinker 18 (10.84%) 10 (6.02%) Current drinking 9 (5.42%) 9 (5.42%) Lifestyle, n (%) 0.676 Basic regular 126 (75.90%) 116 (69.88%) Highly regular 27 (16.27%) 33 (19.88%) Irregular 13 (7.83%) 17 (10.24%) Self-Rated Health Status, n (%) 0.167 poor 96 (57.83%) 90 (54.22%) regular 49 (29.52%) 63 (37.95%) Good 21 (12.65%) 13 (7.83%) Chronic Diseases and Health Conditions Coronary heart disease, n (%) 65 (39.16%) 58 (34.94%) 0.728 Hypertension, n (%) 98 (59.04%) 104 (62.65%) 0.506 Osteoporosis, n (%) 28 (16.87%) 52 (31.33%) 0.008 Stroke, n (%) 35 (21.08%) 47 (28.31%) 0.302 Diabetes, n (%) 61 (36.75%) 59 (35.54%) 0.972 Cataract, n (%) 24 (14.46%) 26 (15.66%) 0.723 Visual Impairment, n (%) 133 (80.12%) 147 (88.55%) 0.035 Vision Problems Affecting Daily Activities, n (%) 27 (16.27%) 40 (24.10%) 0.035 Hearing Impairment, n (%) 70 (42.17%) 73 (43.98%) 0.740 Nutritional and Physical Status MNA Screening Result, n (%) 0.540 Well-nourished 3 (1.81%) 2 (1.20%) At risk of malnutrition 55 (33.13%) 46 (27.71%) Malnutrition 108 (65.06%) 116 (69.88%) Gastritis or Ulcer, n (%) 34 (20.48%) 21 (12.65%) 0.145 Waist Circumference(cm), mean ± SD 87.31 ± 9.72 86.81 ± 11.91 0.028 BMI (kg/m²), n (%) 0.452 < 18.5 2 (1.20%) 17 (10.24%) 18.5–24.9 100 (60.24%) 87 (52.41%) 25–29.9 52 (31.33%) 59 (35.54%) ≥ 30 3 (1.81%) 3 (1.81%) Sleep and Mental Health PSQI Score, mean ± SD 10.45 ± 5.40 9.47 ± 5.35 0.092 Sleep Duration (hours), n (%) 0.058 7 42 (25.30%) 51 (30.72%) Sleep Latency (minutes), n (%) 0.030 ≤ 15 55 (33.13%) 77 (46.39%) 16~30 33 (19.88%) 24 (14.46%) 31~60 20 (12.05%) 25 (15.06%) ≥ 60 58 (34.94%) 40 (24.10%) GDS-5 (depression), n (%) 58 (34.94%) 43 (25.90%) 0.074 GAD-7 (anxiety), n (%) 57 (34.34%) 32 (19.28%) MMSE Score, mean ± SD 22.79 ± 4.53 21.83 ± 5.07 0.094 ADL Score, mean ± SD 96.42 ± 4.57 94.55 ± 5.16 0.001 Chronic Body Pain, n (%) 81 (48.80%) 77 (46.39%) 0.661 Multi-Morbidity and Frailty Multiple Chronic Conditions (≥2), n (%) 146 (87.95%) 154 (92.77%) 0.137 Frailty Status, n (%) 0.165 Pre-frailty, n (%) 115 (69.28%) 103 (62.05%) Frailty, n (%) 51 (30.72%) 63 (37.95%) Limbs Flutter, n (%) 17 (10.24%) 26 (15.66%) 0.141 Urinary Incontinence, n (%) 25 (15.06%) 40 (24.10%) 0.038 Footnote (for both tables): Data are presented as mean ± standard deviation (SD) for continuous variables and n (%) for categorical variables. Between-group differences were assessed using independent-samples t-test for normally distributed continuous variables, Mann–Whitney U test for non-normally distributed continuous variables, and chi-square test (or Fisher’s exact test when appropriate) for categorical variables. P-values < 0.05 were considered statistically significant Table 2. Comparison of Feature Selection Stability Between Full and Balanced Datasets Feature Total Score (Full Dataset) Total Score (Balanced dataset) Overlap Vision_Act 6 6 √ Urine_Incont 6 5 √* Sleep_Time 6 6 √ Sleep_Duration 6 6 √ PSQI_Score 6 6 √ Fried_Stg 6 6 √ FH1lnjury 6 6 √ Cataract 6 6 √ BMI 6 6 √ ADL Score 6 6 √ Vision_Def 5 5 √ Stroke 5 5 √ Smoke_Stat 5 5 √ Self_Hlth 5 4 √* Osteoporosis 5 5 √ Notes： √ indicates that the feature was consistently selected in both datasets with identical total scores. √* indicates that the feature was selected in both datasets but with different total scores, suggesting stable yet slightly variable importance. Table 3 . Performance of Machine Learning Algorithms Across Feature Proportions in Full and Balanced Datasets (Evaluated by AUC-ROC and F1-score) Dataset Feature Proportion Algorithm AUC-ROC F1-score Full Datasets Top 100% Logistic Regression 1 0.98 Top 100% Nearest Neighbors 0.61 0.16 Top 100% Support Vectors 0.95 0.23 Top 100% Decision Tree 1 1 Top 100% Random Forest 0.98 0.75 Top 100% AdaBoost 1 1 Top 100% Gradient Boosting 1 1 Top 100% Naive Bayes 1 0.98 Top 100% Linear DA 1 0.84 Top 100% Quadratic DA 1 1 Top 100% Neural Net 1 0.87 Top 20% Logistic Regression 1 0.98 Top 20% Nearest Neighbors 0.6 0.03 Top 20% Support Vectors 0.94 0.23 Top 20% Decision Tree 1 1 Top 20% Random Forest 1 0.97 Top 20% AdaBoost 1 1 Top 20% Gradient Boosting 1 1 Top 20% Naive Bayes 1 1 Top 20% Linear DA 1 0.84 Top 20% Quadratic DA 1 1 Top 20% Neural Net 1 0.8 Top 10% Logistic Regression 0.62 0 Top 10% Nearest Neighbors 0.47 0.03 Top 10% Support Vectors 0.55 0 Top 10% Decision Tree 0.46 0.11 Top 10% Random Forest 0.52 0.17 Top 10% AdaBoost 0.59 0.07 Top 10% Gradient Boosting 0.54 0.16 Top 10% Naive Bayes 0.62 0.21 Top 10% Linear DA 0.62 0 Top 10% Quadratic DA 0.61 0.16 Top 10% Neural Net 0.62 0 Balanced Datasets Top 100% Logistic Regression 0.99 0.95 Top 100% Nearest Neighbors 0.67 0.45 Top 100% Support Vectors 0.84 0.78 Top 100% Decision Tree 0.99 0.99 Top 100% Random Forest 0.98 0.92 Top 100% AdaBoost 0.99 0.99 Top 100% Gradient Boosting 0.99 0.99 Top 100% Naive Bayes 1 0.97 Top 100% Linear DA 0.89 0.79 Top 100% Quadratic DA 0.99 0.93 Top 100% Neural Net 0.96 0.87 Top 20% Logistic Regression 1 0.99 Top 20% Nearest Neighbors 0.85 0.75 Top 20% Support Vectors 0.98 0.85 Top 20% Decision Tree 0.99 0.99 Top 20% Random Forest 1 0.99 Top 20% AdaBoost 0.99 0.99 Top 20% Gradient Boosting 0.99 0.99 Top 20% Naive Bayes 1 0.99 Top 20% Linear DA 0.99 0.83 Top 20% Quadratic DA 0.61 0.74 Top 20% Neural Net 1 0.98 Top 10% Logistic Regression 0.53 0.56 Top 10% Nearest Neighbors 0.49 0.26 Top 10% Support Vectors 0.5 0.5 Top 10% Decision Tree 0.54 0.48 Top 10% Random Forest 0.47 0.47 Top 10% AdaBoost 0.54 0.57 Top 10% Gradient Boosting 0.5 0.57 Top 10% Naive Bayes 0.52 0.49 Top 10% Linear DA 0.53 0.58 Top 10% Quadratic DA 0.52 0.57 Top 10% Neural Net 0.53 0.57 Additional Declarations No competing interests reported. Supplementary Files SupplementaryMaterial2.docx SupplementaryMaterial1.docx SupplementaryMaterial3.docx SupplementaryMaterial4.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 05 Feb, 2026 Reviews received at journal 29 Dec, 2025 Reviews received at journal 23 Dec, 2025 Reviewers agreed at journal 14 Dec, 2025 Reviewers agreed at journal 12 Dec, 2025 Reviewers invited by journal 12 Dec, 2025 Editor invited by journal 17 Nov, 2025 Editor assigned by journal 14 Nov, 2025 Submission checks completed at journal 14 Nov, 2025 First submitted to journal 03 Nov, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8022235","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":561673007,"identity":"a27252be-4b90-4ddc-9508-c327033e9dad","order_by":0,"name":"Lihua Chen","email":"","orcid":"","institution":"First Affiliated Hospital of Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lihua","middleName":"","lastName":"Chen","suffix":""},{"id":561673008,"identity":"fc6c2947-9055-4365-ade5-97bbceccb66c","order_by":1,"name":"Meiwei Zhang","email":"","orcid":"","institution":"Chongqing University","correspondingAuthor":false,"prefix":"","firstName":"Meiwei","middleName":"","lastName":"Zhang","suffix":""},{"id":561673010,"identity":"58d8c283-20e3-4dc6-a57c-3d9613881b00","order_by":2,"name":"Weihua Yu","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Weihua","middleName":"","lastName":"Yu","suffix":""},{"id":561673014,"identity":"1e764d9b-8dee-476f-a8f1-e4d699892735","order_by":3,"name":"Xintong Liu","email":"","orcid":"","institution":"First Affiliated Hospital of Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xintong","middleName":"","lastName":"Liu","suffix":""},{"id":561673016,"identity":"ecee1ce7-3760-410b-895b-c6c5c8d265d1","order_by":4,"name":"Yang Lü","email":"data:image/png;base64,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","orcid":"","institution":"First Affiliated Hospital of Chongqing Medical University","correspondingAuthor":true,"prefix":"","firstName":"Yang","middleName":"","lastName":"Lü","suffix":""}],"badges":[],"createdAt":"2025-11-03 19:38:25","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8022235/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8022235/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98522699,"identity":"8004af34-23f8-423e-87e0-d5fa242d763e","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"tif","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20904364,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/2b5aec823d5dce0e88cafc28.tif"},{"id":98522697,"identity":"13d4997a-f0da-48c6-aefe-5d4d0a6dfd79","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3541732,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/19ab907b393201208655a44e.docx"},{"id":98522702,"identity":"7a510ea4-d72b-4b5c-aed6-f184441c77c0","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"tif","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29697848,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/cda75e8a72c81033d9fe2371.tif"},{"id":98522711,"identity":"87c7812a-dcb6-4839-ad05-779b6315455f","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"tif","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29927992,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/212f1d37edc09e631ad5e551.tif"},{"id":98522692,"identity":"0f6181c9-ca26-4e76-bb6d-3f2408d580d7","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":36398,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/411ea8b0daa1cfc447a0948d.docx"},{"id":98625969,"identity":"f4bcef8f-8e15-445f-90c3-248bc1cc2477","added_by":"auto","created_at":"2025-12-19 17:09:27","extension":"tif","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16602088,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/c88162ddf161bd6efcac2961.tif"},{"id":98624957,"identity":"c901bbfc-240e-41c0-866f-fcaef9bb023e","added_by":"auto","created_at":"2025-12-19 17:08:51","extension":"docx","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14182,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/41c38caee45465ea37d4ab1b.docx"},{"id":98522728,"identity":"3e2b90e2-c837-481d-92b4-0e7824f4300e","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"tif","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13403440,"visible":true,"origin":"","legend":"","description":"","filename":"Figure5.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/cd0c1ee326dd2aa5313846f7.tif"},{"id":98625006,"identity":"ebf07bdb-0101-4d61-94f6-b01cfbb34641","added_by":"auto","created_at":"2025-12-19 17:08:53","extension":"docx","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22935,"visible":true,"origin":"","legend":"","description":"","filename":"Table3..docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/f24f8341fb9eed7e0c41ade6.docx"},{"id":98625492,"identity":"8332d277-2825-4699-b084-5a3f8913a0d1","added_by":"auto","created_at":"2025-12-19 17:09:10","extension":"json","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7218,"visible":true,"origin":"","legend":"","description":"","filename":"7de64663d6504990a4c9e72ce8d57617.json","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/3c45ad2612e3fec194dc9464.json"},{"id":98624480,"identity":"45d36d1a-98bf-4285-9107-de5f2eca0031","added_by":"auto","created_at":"2025-12-19 17:08:26","extension":"docx","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16714,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/0bfef3c843905a84b6e1b96a.docx"},{"id":98624688,"identity":"d42d5f0b-1a86-4878-b2fc-d8b0d09a5246","added_by":"auto","created_at":"2025-12-19 17:08:39","extension":"docx","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21594,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/083d2d6ce18c488b43344b18.docx"},{"id":98625109,"identity":"185f9f01-0027-4c20-844d-8c9f4239b9e2","added_by":"auto","created_at":"2025-12-19 17:08:56","extension":"docx","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38466,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial3.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/6ec1d683bc1c4bfcff725285.docx"},{"id":98624518,"identity":"cf3f7b4b-f252-4b92-8b31-dfe2ab890db2","added_by":"auto","created_at":"2025-12-19 17:08:29","extension":"docx","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":112777,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial4.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/0947d781a8116540ab85f6bd.docx"},{"id":98522705,"identity":"07b045dd-f60c-4c67-b379-4668390671fa","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"xml","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":275864,"visible":true,"origin":"","legend":"","description":"","filename":"7de64663d6504990a4c9e72ce8d576171enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/73e6f8b591633f946834588c.xml"},{"id":98624420,"identity":"925d1038-b296-4b4d-bcdf-d78e1a5a9b62","added_by":"auto","created_at":"2025-12-19 17:08:24","extension":"tif","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20904364,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/36790cc42c5448fbc7083f66.tif"},{"id":98522729,"identity":"b44887c0-5485-4d86-b392-2fbdc26257ca","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"tif","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29697848,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/bb4c0107ad8fe40488869c86.tif"},{"id":98522718,"identity":"67919862-022f-4814-98ff-8b2a8dfb1155","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"tif","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29927992,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/9542a300573f13eda8f3c570.tif"},{"id":98624549,"identity":"cd625339-2951-4b8a-aa07-2d36e30f8b7a","added_by":"auto","created_at":"2025-12-19 17:08:30","extension":"tif","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16602088,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/998c5e04fe2f99bae3b11a2d.tif"},{"id":98522716,"identity":"77f17c5c-593d-478b-b3c4-f50d54132f11","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"tif","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13403440,"visible":true,"origin":"","legend":"","description":"","filename":"Figure5.tif","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/2b4b9053d38d31caa0b55913.tif"},{"id":98522703,"identity":"75616dd6-4ad4-4908-83c7-c668d76c5ee4","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"jpeg","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":856816,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/442f901271cc42083d9d54c0.jpeg"},{"id":98625256,"identity":"3cf81516-c190-4349-a470-cf71592b2545","added_by":"auto","created_at":"2025-12-19 17:09:00","extension":"jpeg","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1783250,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/eb9a0920dd6025ca7986ad15.jpeg"},{"id":98522710,"identity":"10307ee7-bdb5-468c-ab1c-8304873fe1e0","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1029774,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/54b05526e110b75583316c39.png"},{"id":98624425,"identity":"9781e6e8-5ef0-4761-a724-60675a5d1907","added_by":"auto","created_at":"2025-12-19 17:08:24","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":577906,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/71d82b85ca8814845490a707.png"},{"id":98625567,"identity":"980fea6f-4bc5-4efc-8965-ba28defbb765","added_by":"auto","created_at":"2025-12-19 17:09:11","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":231026,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/bc78bc7e2bcfc1526717ba9b.png"},{"id":98625018,"identity":"5d820df0-0f9f-4162-bf91-b457565ada1a","added_by":"auto","created_at":"2025-12-19 17:08:53","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":206986,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/83e5098e310f02bb38510cdc.png"},{"id":98625450,"identity":"4a0c8c81-e473-40a2-964f-95d5a646e20f","added_by":"auto","created_at":"2025-12-19 17:09:08","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3169439,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/a39a9d3f6117afd40a00e433.png"},{"id":98522732,"identity":"872c49e0-7d6b-44c5-8126-c5c1200b76b5","added_by":"auto","created_at":"2025-12-18 13:57:29","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3296683,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/41d34dc9fc2e62c45be3da5e.png"},{"id":98625015,"identity":"2e640f2d-933a-4440-b713-8581228625fd","added_by":"auto","created_at":"2025-12-19 17:08:53","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1204878,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/9712cd64e30568abb3c00022.png"},{"id":98522720,"identity":"7fe2c715-4a6a-4163-ab29-9b69594cacd8","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":521791,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/ba9dfbee54883bf1eda3a554.png"},{"id":98522713,"identity":"0ec43011-d2ae-4516-b37d-a3591d54d954","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"png","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":184483,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/66e4b698000dc28f88388e8a.png"},{"id":98624691,"identity":"b05e6334-bb08-4ceb-a581-330417d31004","added_by":"auto","created_at":"2025-12-19 17:08:39","extension":"png","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":393583,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/31cb85b0fe628b89ee202371.png"},{"id":98522724,"identity":"cb37d059-4ccb-4edc-843b-c6d19db15b77","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"png","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":207379,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/e82de393d429ddb0b0da62eb.png"},{"id":98522721,"identity":"b8b3479b-8263-4e71-af02-6f90bb7019f8","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"png","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":114394,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/ca32cb02383537faf3b1a78f.png"},{"id":98625067,"identity":"fed2890c-37de-4444-8ca6-42fb67f58c36","added_by":"auto","created_at":"2025-12-19 17:08:54","extension":"png","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49922,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/03b5295ba03d8e419dc62092.png"},{"id":98522714,"identity":"6a09a7d6-bf15-4d0d-81ca-51a75367028f","added_by":"auto","created_at":"2025-12-18 13:57:28","extension":"xml","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":281404,"visible":true,"origin":"","legend":"","description":"","filename":"7de64663d6504990a4c9e72ce8d576171structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/ef1034ca838c8470629f60ca.xml"},{"id":98625473,"identity":"c2520b48-3ed6-45ff-a16c-b3e980a7d91f","added_by":"auto","created_at":"2025-12-19 17:09:09","extension":"html","order_by":36,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":294343,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/00bbb8f6c93ae37eade43367.html"},{"id":98625989,"identity":"4919c043-33bb-45d1-a117-efddf1c0e6ac","added_by":"auto","created_at":"2025-12-19 17:09:27","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":4627981,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe overall flowchart of the study\u003c/strong\u003e. (A) The algorithm chart of the study. (B) Flow of study participants through the study.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/376d281dee1391c3411b3425.png"},{"id":98624647,"identity":"95df8364-3d48-4577-a1af-db77d5f16006","added_by":"auto","created_at":"2025-12-19 17:08:37","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":10275708,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves for machine learning models in the full dataset\u003c/p\u003e\n\u003cp\u003ePanel A: the AUC-ROC values represent the performance of prediction models developed utilizing the top 100% to the top 10% of features (panel A :subfigures a–j in the full dataset).\u003c/p\u003e\n\u003cp\u003ePanel B: the AUC-ROC values reflect the performance of prediction models constructed with the bottom 100% to the bottom 10% of features (panel B: subfigures a–j in the full dataset).\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/875cc1d8dccea9b9b190a25e.png"},{"id":98625261,"identity":"d3af7a85-5d21-4192-af0f-2b160c0ff68c","added_by":"auto","created_at":"2025-12-19 17:09:00","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":10573940,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves for machine learning models in the Balanced dataset\u003c/p\u003e\n\u003cp\u003ePanel A: the AUC-ROC values represent the performance of prediction models developed utilizing the top 100% to the top 10% of features (panel A :subfigures a–j in the balanced dataset).\u003c/p\u003e\n\u003cp\u003ePanel B: the AUC-ROC values reflect the performance of prediction models constructed with the bottom 100% to the bottom 10% of features (panel B: subfigures a–j in the balanced dataset).\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/11a65d1a61a8b0851551b61e.png"},{"id":98625071,"identity":"a6b5770b-787e-4576-81d6-4cef3ee56379","added_by":"auto","created_at":"2025-12-19 17:08:54","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":3567195,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of machine learning algorithms across feature subsets in full and balanced datasets.\u003c/p\u003e\n\u003cp\u003e(a) Line plots of AUC-ROC values for 11 machine learning algorithms in the full dataset, evaluated across top (100%–10%) and bottom (90%–10%) feature subsets.\u003c/p\u003e\n\u003cp\u003e(b) Corresponding bar plots of F1-scores in the full dataset under the top 100%, 50%, 20%, and 10% feature subsets, as well as the bottom 90%.\u003c/p\u003e\n\u003cp\u003e(c) Line plots of AUC-ROC values in the balanced dataset, across the same feature subsets.\u003c/p\u003e\n\u003cp\u003e(d) Corresponding bar plots of F1-scores in the balanced dataset under identical feature subset conditions.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/b08433ec836b4248bf59e566.png"},{"id":98522695,"identity":"4782400c-f0c0-4e9f-b0ff-fe9cb98d4850","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1448093,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance rankings of predictors for fall risk derived from random forest models in the full (Panel A) and balanced (Panel B) datasets.\u003c/p\u003e\n\u003cp\u003ePanel A (Full Dataset): Feature importance scores estimated from the random forest model using the complete dataset. Panel B (Balanced Dataset): Feature importance scores estimated from the random forest model after under-sampling to balance the outcome classes.\u003c/p\u003e\n\u003cp\u003eX-axis: Feature importance values (mean decrease in Gini impurity).\u003c/p\u003e\n\u003cp\u003eY-axis: Candidate predictors ordered by importance.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/d15661f67c82d16b8ae22306.png"},{"id":98774908,"identity":"963080ff-ec61-4932-a0f3-591be142afb8","added_by":"auto","created_at":"2025-12-22 12:16:39","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":31613593,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/2dc5ef2a-bd6b-4424-a537-3adf7f3bd839.pdf"},{"id":98522688,"identity":"374f0d1f-b86c-4b32-b83b-ffdbe430f3dc","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":21594,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/528f918862ba32321bcaab45.docx"},{"id":98522687,"identity":"125e9673-512a-4277-9ee2-2f6830bfefe1","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":16714,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/5da144e0acb87ccf9990df61.docx"},{"id":98625413,"identity":"1ed34aab-8589-4a41-af8f-87cd8461ef9f","added_by":"auto","created_at":"2025-12-19 17:09:06","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":38466,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial3.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/83871fa811b2705ee42b0c36.docx"},{"id":98522696,"identity":"d622a140-a3eb-463b-b37d-cb5543f1ab71","added_by":"auto","created_at":"2025-12-18 13:57:27","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":112777,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial4.docx","url":"https://assets-eu.researchsquare.com/files/rs-8022235/v1/5fcb3b915cb312a8df089eb8.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Risk Prediction Models for Falls Among Older Adults Inpatients with Cognitive frailty: Machine Learning Study Based on Comprehensive Geriatric Assessment","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eThe rapid aging of the global population has led to an increased prevalence of age-related health issues, particularly falls, which are among the most common and preventable causes of injury-related morbidity and mortality for old adults [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAccording to the World Health Organization, an estimated 684,000 people die each year from falls worldwide [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Studies have shown that individuals over the age of 60 account for the highest proportion of fatal falls. Specifically, the incidence of falls ranges from 28% and 35% among those aged 65, and from 32% to 42% among individuals over 70 years of age [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Falls can lead to significant physical injuries, including fractures\u0026mdash;especially hip fractures [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], head traumas, and even paralysis. These injuries can result in impaired physical function, prolonged rehabilitation periods and a diminished quality of life. Moreover, the financial burden of falls in older adults is substantial, with costs estimated to range from approximately \u003cspan\u003e$\u003c/span\u003e2,044 to \u003cspan\u003e$\u003c/span\u003e25,955 [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. This highlights the urgent need for the early identification of individuals at risk for falls, which is essential for reducing fall incidents and improving overall health outcomes.\u003c/p\u003e \u003cp\u003eAlthough the high incidence and severe outcomes of falls have been extensively documented, it is crucial to focus on the vulnerabilities of specific populations who are more prone to such incidents. Cognitive frailty refers to a combined condition characterized by the simultaneous presence of both physical frailty and cognitive impairment. The concept of cognitive frailty was introduced by an international consensus group from the International Academy of Nutrition and Aging (IANA) in collaboration with the International Association of Gerontology and Geriatrics (IAGG) in 2013. This highlighted the need for targeted interventions in this population [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Individuals with cognitive frailty had a 45% higher risk of falling compared to those without cognitive impairment [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Research has demonstrated that patients with cognitive frailty face greater challenges in walking and maintaining balance due to factors such as reduced muscle strength, diminished physical ability, impaired cognitive function, poor environmental adaptability, and social isolation [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Therefore, timely identification and intervention regarding fall risks in this population are essential.\u003c/p\u003e \u003cp\u003eDespite the availability of various clinical assessment tools, such as the Tinetti Balance Assessment and Berg Balance Scale, these tools have limited predictive power in cognitively frail populations. They often fail to account for the multifaceted nature of fall risk in these individuals, which includes not only physical but also cognitive, nutritional, and psychological factors. Additionally, cultural nuances and socio-environmental contexts prevalent in Asian societies may further influence fall risk profiles and the effectiveness of generalized assessment tools. Consequently, there is a compelling need for more accurate and personalized predictive models specifically tailored for Asian older adults with cognitive frailty to address the shortcomings of existing tools in this context.\u003c/p\u003e \u003cp\u003eThe adoption of machine-learning (ML) models in clinical environments has gained significant attention due to its ability to formulate robust risk models and enhance predictive performance [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. ML has emerged as a compelling alternative to traditional multiple linear regression (MLR) in medical research due to its ability to capture non-linear relationships and intricate interactions among numerous predictors without the assumption of a normal data distribution [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. As a result, ML can potentially outperform conventional MLR in disease prediction [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. While various machine learning methodologies for fall risk prediction have been explored in previous studies, models specifically targeting older adults with cognitive frailty, particularly within Asian populations, remains relatively under-investigated.\u003c/p\u003e \u003cp\u003eTo address these gaps, this study aims to develop a series of machine learning models to accurately predict the risk of falls in Asian older adult inpatients with cognitive frailty, utilizing data from Comprehensive Geriatric Assessment (CGA). By integrating a diverse array of variables encompassing cognitive function, physical frailty, nutritional status, and psychological well-being, we seek to enhance the precision of fall risk prediction and to identify critical determinants contributing to falls in this specific demographic. The resulting models will provide clinicians with effective prospective decision support tools, facilitating the timely and accurate identification of high-risk individuals for targeted early interventions, thereby reducing fall incidence and improving patient outcomes. This research holds significant practical implications for advancing geriatric healthcare management in Asian regions and addressing the challenges posed by population aging.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eData source\u003c/h2\u003e\n \u003cdiv id=\"Sec4\" class=\"Section3\"\u003e\n \u003ch2\u003eData collection\u003c/h2\u003e\n \u003cp\u003eThe datasets used for constructing the ML model were derived from the CGA Database of hospitalized patients in the geriatric department at The First Affiliated Hospital of Chongqing Medical University, covering the period from March 2016 to June 2023. Inclusion criteria were: (1) Age\u0026thinsp;\u0026ge;\u0026thinsp;60 years; (2) Complete assessment data available; (3) For patients with multiple assessments, only data from the first assessment were used. The excluded criteria were: (1) patients with only physical frailty, only cognitive impairment, or dementia; (2) patients under the age of 60 years; (3) patients with essential information missed or over 20% of the information lost.\u003c/p\u003e\n \u003cp\u003eAll participants underwent in-person assessments by professional evaluators. The evaluator held a postgraduate degree in nursing, was qualified as a comprehensive geriatric assessor through training, and had over twenty years of experience. The research methodology adheres to the principles of the Declaration of Helsinki and its revisions, as approved by the ethics board of The First Affiliated Hospital of Chongqing Medical University. Informed consent was waived for this retrospective study, which follows STROBE guidelines. An overview of the study workflow is presented in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003ch3\u003eDefinition and Assessment of Cognitive Frailty\u003c/h3\u003e\n\u003cp\u003eThe diagnosis of cognitive frailty is based on the diagnostic criteria proposed by the International Academy on Nutrition and Aging (IANA) and the International Association of Gerontology and Geriatrics (IAGG) in 2013: (i) the presence of both physical frailty and cognitive impairment, with a Clinical Dementia Rating (CDR) of 0.5; and (ii) the exclusion of Alzheimer\u0026rsquo;s disease or other types of dementia. In our study, cognitive frailty is characterized by the concurrent presence of at least one of Fried\u0026apos;s criteria and mild cognitive impairment as defined by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5)\u003c/p\u003e\n\u003ch3\u003eAssessment of Physical Frailty\u003c/h3\u003e\n\u003cp\u003ePhysical frailty is defined using Fried\u0026rsquo;s criteria, which involve five components: (i) unintentional weight loss, characterized by an involuntary loss of more than 5 kg in the past year; (ii) exhaustion, assessed through two self-reported questions from the Center for Epidemiological Studies Depression Scale: \u0026ldquo;How many times in the past week have you struggled to get anything done?\u0026rdquo; and \u0026ldquo;How often do you feel unable to move forward?\u0026rdquo; A response of \u0026ldquo;often\u0026rdquo; or \u0026ldquo;most of the time\u0026rdquo; to either question indicates exhaustion; (iii) weak muscle strength, evaluated by handgrip strength using an electronic hand dynamometer (Zhongshan Camry Electronic Co. Ltd, Guangdong, China). Participants stand and grip the dynamometer with maximum force using their dominant hand, with two trials conducted to record the maximum value. Grip strength less than 28 kg for men or less than 18 kg for women indicates reduced strength; (iv) slowness, assessed via a 6-meter fast gait speed test, where a speed of less than 1.0 m/s signifies slowness; and (v) low physical activity, determined by the question \u0026ldquo;How do you usually engage in physical exercise?\u0026rdquo; Responses such as \u0026ldquo;no physical exercise\u0026rdquo; or \u0026ldquo;mostly sedentary\u0026rdquo; indicate low activity levels.\u003c/p\u003e\n\u003cp\u003eEach of the aforementioned criteria contributes one point. Participants are categorized as frail (three or more points), pre-frail (one or two points), or robust (zero points). To enhance the recognition of cognitive frailty, individuals in pre-frail states are also included in our analysis.\u003c/p\u003e\n\u003ch3\u003eAssessment of Cognitive Function\u003c/h3\u003e\n\u003cp\u003eCognitive function was assessed using the Mini-Mental State Examination (MMSE), which provides a comprehensive and accurate evaluation of a subject\u0026rsquo;s intellectual status and degree of cognitive impairment. The MMSE encompasses seven dimensions: orientation to time, orientation to place, immediate memory, delayed memory, attention and calculation, language, and visual-spatial abilities. The MMSE is a 30-item scale with scores ranging from 0 to 30, where a higher score indicates better cognitive function.\u003c/p\u003e\n\u003cp\u003eParticipants with intact cognition are defined as those with an MMSE score of 27 or greater and no clinical diagnosis of cognitive impairment. For patients with mild cognitive impairment (MCI), the diagnostic criteria are based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) by the American Psychiatric Association (2013): (i) subjective and objective examinations indicating MCI; (ii) cognitive decline in one or more of the aforementioned dimensions; (iii) preserved daily living abilities; (iv) absence of dementia diagnosis; (v) exclusion of other conditions causing cognitive decline; and (vi) MMSE scores. The MMSE scores are adjusted based on education level, with cut-off points of 17 for illiterate individuals, 20 for primary education, 22 for junior high education, and 23 for university-level or higher education.\u003c/p\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003eOutcome Measure\u003c/h2\u003e\n \u003cp\u003eThe primary outcome measure was the occurrence of falls, defined by the variable \u0026ldquo;Fall History\u0026rdquo; (FH), which indicates whether a participant experienced a fall in the past two years. Additionally, the variable \u0026ldquo;First-Ever Fall-Related Injury\u0026rdquo; (FH1-Injury), a binary indicator (Yes/No), captured whether the participant\u0026rsquo;s first fall resulted in an injury requiring medical attention. This variable reflects the historical severity of the participant\u0026rsquo;s first lifetime injurious fall, which could have occurred long before the observation window for FH. Importantly, FH1-Injury represents the first-ever injurious fall, making it temporally independent of the study\u0026rsquo;s observation period for recent falls.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003ePredictive Variables\u003c/h3\u003e\n\u003cp\u003eIn this study, 93 candidate predictor variables were initially considered, including demographic factors (age, gender, lifestyle), clinical indicators (MMSE, PSQI), and functional characteristics (vision impairment, ADL scores). To address potential redundancy and multicollinearity, Principal Component Analysis (PCA) was applied, reducing dimensionality and creating composite features such as PCA_Hip_Circ_Waist_Circ and PCA_Hear_Act_Hear_Def. Pearson correlation analysis revealed significant correlations between variables (e.g., Waist_Circ and Hip_Circ: 0.731; Hear_Act and Hear_Def: 0.89). After preprocessing, 58 variables were retained for final analysis. A detailed list and descriptions of all predictor variables are provided in Supplementary Material 1.\u003c/p\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n \u003ch2\u003eStatistical Analysis\u003c/h2\u003e\n \u003cp\u003eContinuous variables were shown as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD with range or median with IQR, based on distribution. Group comparisons used the Student\u0026rsquo;s t-test for normal data or Wilcoxon rank-sum test for non-normal data. Categorical variables were presented as frequencies and percentages, with comparisons using the chi-square test or Fisher\u0026rsquo;s exact test. Descriptive and variance analyses were conducted with SPSS Version 26.0, while data preprocessing, feature selection, ML model development, and evaluation were carried out using Python 3.8. Statistical significance was set at p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003eData processing\u003c/h2\u003e\n \u003cp\u003eThe overall workflow of data preprocessing, feature selection, model construction, and performance evaluation is illustrated in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eA. Specifically, data preprocessing involved addressing missing values, normalizing continuous variables, and encoding categorical variables. To balance the number of fall and non-fall cases, random under-sampling was applied. Six feature selection methods\u0026mdash;Pearson Correlation, Chi-square test, Recursive Feature Elimination (RFE), Logistic Regression, Random Forest, and Light Gradient Boosting Machine (Light GBM)\u0026mdash;were utilized to identify the most relevant predictors of fall risk. Feature selection is performed on six feature sequences, F1, F2, F3, F4, F5, and F6 (F1-F6 refer to the ordered sequence of features in the different selection methods, more detailed steps were seen in the Supplementary Material 2, section A), followed by model construction. The final feature sequence F:\u003c/p\u003e\n \u003cdiv id=\"Equa\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e$$\\:F=\\sum\\:_{i=1}^{i=6}{W}_{i}{F}_{i}$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eHere, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{W}_{i}\\)\u003c/span\u003e\u003c/span\u003e represents the weight parameter corresponding to the i-th feature analysis method, taking a value of 0 or 1, where 1 indicates retention of the feature, and 0 denotes its exclusion. After feature selection, 58 features were retained, and the features were subsequently grouped based on their importance rankings for further analysis.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003eAlgorithms\u003c/h2\u003e\n \u003cp\u003eA total of eleven ML algorithms were employed to build predictive models for each feature subset. The algorithms used included Decision Tree, Gradient Boosting, Quadratic Discriminant Analysis (QDA), AdaBoost, Naive Bayes, Logistic Regression, Random Forest, Linear Discriminant Analysis (LDA), Neural Network, and Support Vector Machines (SVM), as well as k-Nearest Neighbors (k-NN). Each machine learning algorithm was applied to every feature subset, and the results from all datasets were compiled for comparison and analysis. Detailed procedures for model development are presented in Supplementary Material 2, Section B, and the parameter tuning strategies for the machine learning algorithms are described in Supplementary Material 2, Section C table.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003eModel evaluation indexes\u003c/h2\u003e\n \u003cp\u003eThe primary metric for model performance evaluation was the area under the receiver operating characteristic curve (AUC-ROC). In addition, various performance metrics including accuracy, sensitivity, specificity, recall, precision, and F1 score were also examined. All statistical analyses and modeling procedures were conducted using Python (version 3.8), with statistical significance set at a two-sided P-value of less than 0.05. To avoid over-fitting and to provide an unbiased estimate of generalisation performance, we adopted a nested, repeated 10-fold cross-validation (CV). A two-stage procedure was used:\u003c/p\u003e\u003cspan\u003e\n \u003cp\u003e\u003cstrong\u003e(i) outer loop (model assessment)\u003c/strong\u003e\u0026ensp;\u0026ndash;\u0026ensp;a stratified 10-fold CV repeated \u003cstrong\u003ethree\u003c/strong\u003e times (30 train/test splits in total) delivered point estimates and 95% CIs of each metric;\u003c/p\u003e\n \u003c/span\u003e \u003cspan\u003e\n \u003cp\u003e\u003cstrong\u003e(ii) inner loop (model selection)\u003c/strong\u003e\u0026ensp;\u0026ndash;\u0026ensp;for every training partition in the outer loop, a further stratified 10-fold CV performed random/grid search over the hyper-parameter space. Feature imputation, scaling, and the feature-selection step were treated as components of a single Pipeline object and re-computed inside the inner loop to prevent information leakage. The tuning objective was ROC-AUC; ties were broken by favouring the more parsimonious model (smaller tree depth, fewer estimators, or lower C). All experiments were implemented with scikit-learn 1.5.0, imbalanced-learn 0.12.0, and NumPy 1.26; random seed fixed at 42 for full reproducibility.\u003c/p\u003e\n \u003c/span\u003e\n \u003cp\u003eA detailed description of model performance evaluation is presented in Supplementary Material 2, Section D.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003e1. Demographic and Clinical Characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 814 participants were included in the study based on the inclusion criteria. To address the significant sample size imbalance between the fallers and non-fallers, we applied a random under-sampling technique, resulting in the exclusion of 432 participants (Figure 1B). The initial dataset is referred to as the \u0026quot;full dataset\u0026quot; (n = 814), while the dataset after under-sampling is referred to as the \u0026quot;Under-sampled Balanced Dataset \u0026quot; (n = 332).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.1 Full Dataset (n = 814)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs shown in Table 1A, the demographic and clinical characteristics of fallers and non-fallers in the full dataset demonstrated significant group differences. Fallers (n = 166) were significantly older than non-fallers (n = 648), with a mean age of 78.05 \u0026plusmn; 7.31 years versus 76.32 \u0026plusmn; 7.92 years, respectively (\u003cem\u003eP\u003c/em\u003e = 0.006). Other demographic characteristics, such as gender and educational level, did not show significant differences between the two groups (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05). Notably, fallers had higher rates of osteoporosis (31.3% vs. 21.1%, \u003cem\u003eP\u003c/em\u003e = 0.013) and visual impairment (88.6% vs. 79.0%, \u003cem\u003eP\u003c/em\u003e = 0.005) compared to non-fallers. Additionally, fallers were more likely to have vision problems affecting daily activities (24.1% vs. 12.96%, \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001) and urinary incontinence (24.1% vs. 15.4%, \u003cem\u003eP\u003c/em\u003e = 0.008).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.2 Under-sampled Balanced Dataset (n = 332)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCorrespondingly, Table 1B presents the demographic characteristics in the balanced dataset,\u0026nbsp;showing a similar pattern of group differences after random under-sampling. Fallers were also significantly older (\u003cem\u003eP\u003c/em\u003e = 0.028) and had higher rates of osteoporosis (31.3% vs. 16.9%, \u003cem\u003eP\u003c/em\u003e = 0.008) and visual impairment (88.6% vs. 80.1%, \u003cem\u003eP\u003c/em\u003e = 0.035). Urinary incontinence remained a distinguishing factor between fallers and non-fallers (24.1% vs. 15.1%, \u003cem\u003eP\u003c/em\u003e = 0.038). However, differences in smoking status and gender distribution were less pronounced in this dataset compared to the full sample.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2. Key Predictors of Fall Risk\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAcross both the fully sampled dataset (n = 814) and the under-sampled balanced dataset (n = 332), several features consistently emerged as significant predictors of fall risk. These predictors were identified through feature selection methods (Pearson/\u0026chi;\u0026sup2;, RFE) and model-based importance ranking (logistic regression, random forest, LightGBM). The top-ranked predictors spanned four key domains: vision function (Vision_Act, Vision_Def), sleep characteristics (Sleep_Time, Sleep_Duration, PSQI), functional and cognitive status (lower ADL, slightly lower MMSE), and frailty/multimorbidity (Fried stage, \u0026ge;2 chronic conditions), with urinary incontinence and BMI.\u0026nbsp;Across both datasets, the importance rankings of most features were largely consistent. Table 2 summarizes the scores of the top 15 features identified in both the full and balanced datasets. Dataset-specific differences were minimal: Age and residence area were slightly more relevant in the balanced dataset but did not affect the overall ranking or the direction of associations. Collectively, these findings across different methods and datasets indicate a stable, multifactorial risk profile for falls in cognitively frail older adults. The complete ranking results for all 58 features in both datasets are provided in Supplementary Material 3 (Section A for full dataset, section B for balanced dataset).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3. Model Development and Performance Across Feature Subsets\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe developed fall risk prediction models using 11 machine learning algorithms, evaluated across different feature proportions, ranging from the full feature set (100%) to the top 10%, as well as the bottom 90% to bottom 10% subsets. Figures 2 and 3 present the ROC curves for all machine learning classifiers under different feature subsets in the full and balanced datasets, respectively. In the full dataset (Figure 2), models trained on the top-ranked features (Panel A) consistently achieved high discrimination, with ensemble methods (AdaBoost, Gradient Boosting, Random Forest, Decision Tree) and Logistic Regression yielding AUCs close to 1.00 when\u0026nbsp;\u0026ge;20% of features were included. In contrast, models trained on the bottom-ranked features (Panel B) showed a rapid decline in discriminative ability, with ROC curves approaching the diagonal line as feature subsets decreased, underscoring the limited predictive value of low-ranked variables. Similarly, in the balanced dataset (Figure 3), the top feature subsets (Panel A) sustained strong predictive performance across classifiers, with AUCs consistently \u0026gt;0.90 when at least 20% of features were retained. However, bottom feature subsets (Panel B) again demonstrated markedly reduced performance, with AUCs clustering around 0.5, highlighting the robustness of the top-ranked predictors even under class-balancing conditions.\u003c/p\u003e\n\u003cp\u003eTo further quantify these findings, Figure 4 complements the ROC curves by providing a comparative overview of AUC-ROC values (Panels a, c) and F1-scores (Panels b, d) across top and bottom feature subsets. The line plots clearly show that AUC remained near-perfect when using the top 20% of features, whereas performance declined sharply with bottom-ranked subsets. The corresponding bar plots reinforce this observation, demonstrating that models achieved the highest F1-scores with top features, while bottom subsets led to unstable and substantially lower F1 values.\u003c/p\u003e\n\u003cp\u003eFinally, these graphical patterns are substantiated in Table 3, which summarizes the performance of all 11 algorithms under the full, 20%, and 10% top feature subsets. The performance of all models across every feature proportion (from Top100% to Top10% and Bottom90% to Bottom10%) is provided in Supplementary Material 4. Ensemble methods (AdaBoost, Gradient Boosting, Random Forest) and Decision Trees consistently achieved near-perfect AUC (\u0026asymp;1.00) with the top 100% and 20% features, while Logistic Regression offered comparable accuracy with greater interpretability. However, when restricted to only the top 10% of features, performance variability across algorithms increased, and F1-scores were generally reduced. This integrated analysis across ROC curves, line/bar plots, and tabular results provides convergent evidence that top-ranked features not only preserve predictive accuracy but also yield a more stable and clinically interpretable fall-risk model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4. Feature Importance and Clinical Thresholds\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFigure 5 illustrates the feature importance rankings derived from random forest models after training. Across both the fully sampled (figure 5a) and under-sampled balanced datasets (figure 5b), seven variables consistently emerged as the most influential predictors of fall risk: history of first injurious fall (FH1-Injury), ADL score, age, waist circumference, hearing deficit, GAD-7 score, and MMSE score. The consistent ranking across sampling strategies indicates the stability of these predictors. Additional contributors, such as vision impairment, sleep quality indicators (e.g., PSQI score, sleep duration, latency), and frailty status (e.g., Fried stage), further highlight the multifactorial nature of fall risk in cognitively frail older adults.\u003c/p\u003e\n\u003cp\u003eTo enhance robustness, we performed supplementary analyses using random forest\u0026ndash;based impurity reduction and neural network\u0026ndash;based permutation importance. Both approaches supported the stability of the identified predictors, although minor variability and occasional negative values were observed with permutation importance. Such fluctuations likely reflect methodological sensitivity to noise, overfitting, or multicollinearity. Importantly, the consistency of results across methods provides convergent evidence for the reliability of these core predictors.\u003c/p\u003e\n\u003cp\u003eTo improve interpretability, the direction and threshold effects of key predictors were examined. A lower ADL score (\u0026lt;100) and lower MMSE score (\u0026lt;19) were both associated with higher fall risk, reflecting reduced functional and cognitive capacity. Similarly, advanced age (\u0026ge;79 years) and elevated GAD-7 scores (\u0026ge;1), reflecting greater anxiety symptoms, were associated with a higher likelihood of falls. In contrast, better functional independence, preserved cognitive performance, and psychological stability were protective against fall risk. These threshold patterns remained consistent across analytical models, reinforcing the robustness and reproducibility of these key predictors.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study provides compelling evidence that machine learning (ML) applied to comprehensive geriatric assessment (CGA) data can yield highly accurate and clinically meaningful fall risk predictions in older adults with cognitive frailty within two years. By integrating both full-sampling and under-sampling strategies across multiple ML methods, we identified a robust and reproducible set of seven core predictors of fall risk: history of first injurious fall, activities of daily living (ADL) score, age, waist circumference, hearing deficit, GAD-7 score, and MMSE score. For several of these, we determined clinically actionable threshold values (e.g., ADL\u0026thinsp;=\u0026thinsp;100, age\u0026thinsp;=\u0026thinsp;79 years, MMSE\u0026thinsp;=\u0026thinsp;19) beyond which fall risk markedly increases. These threshold-based insights provide a foundation for individualized risk stratification and targeted intervention in this high-risk population.\u003c/p\u003e \u003cp\u003eKey Predictors and Clinical Relevance: Each of the core predictive features is strongly associated with known biological mechanisms and clinical manifestations of fall risk in cognitively frail older adults. A history of an injurious fall emerged as an especially influential predictor, aligning with findings by Chen et al[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e],. that prior falls are a key risk factor for future falls in community-dwelling elders. An initial injurious fall can lead to functional decline, manifested as reduced muscle strength, impaired balance, and limited joint mobility. It often induces a fear of falling, which in turn reduces physical activity and further impairs gait stability. This vicious cycle increases the risk of subsequent falls and recurrent injuries. The ADL score was another consistently top-ranked predictor; lower ADL scores indicate greater functional dependency and frailty, reflecting declines in physical capacity and independence [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Psychological status, as measured by the GAD-7, was also a strong contributor to the model. Higher anxiety levels were associated with increased fall risk, which is biologically plausible given that anxiety and related affective disturbances can impair attention, psychomotor coordination, and executive function, all of which are essential for safe mobility. This observation is consistent with the report by Wang et al. [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e] linking anxiety and depression to significantly higher risks of falls and fall-related injuries in older adults.\u003c/p\u003e \u003cp\u003eWaist circumference showed a positive association with fall risk in our study. Notably, 7.3% of participants had a waist circumference exceeding 90 cm (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD: 86.7\u0026thinsp;\u0026plusmn;\u0026thinsp;10.8 cm). Although the odds ratio at this threshold was modest and not statistically significant (OR\u0026thinsp;=\u0026thinsp;1.03, 95% CI: 0.73\u0026ndash;1.46), we considered a 90 cm waist as a potentially important cutoff for clinical attention. Central obesity, as indicated by an excessive waistline, is known to impair balance and restrict mobility [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Thus, a waist circumference above 90 cm may serve as a practical marker for elevated fall risk in cognitively frail older adults, though this proposition requires validation in larger cohorts. Hearing deficit also emerged as a key predictor in our models. Sensory impairments such as hearing loss can diminish situational awareness and increase cognitive load during ambulation, hampering an individual\u0026rsquo;s ability to maintain balance and detect environmental hazards. This mechanism is supported by epidemiological studies [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e] and by Gopinath et al. [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], who reported significantly higher fall rates among older adults with multisensory impairments (e.g., concurrent hearing and vision loss).\u003c/p\u003e \u003cp\u003eCognitive deficits contributed significantly to fall risk as well. In addition to hearing loss, global cognitive function (assessed by the MMSE) was a consistent predictor in our analysis. We identified an optimal MMSE cutoff score of 19 (Youden\u0026rsquo;s index\u0026thinsp;=\u0026thinsp;0.027; sensitivity 22.4%, specificity 80.3%), beyond which fall risk rose appreciably. Although the discriminatory power of this specific threshold was modest, it aligns with the notion that even moderate cognitive impairment can heighten fall susceptibility. Consistent with our findings, a previous study [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e],found that cognitive decline in older adults was significantly associated with an increased incidence of serious fall-related injuries. These observations underscore the importance of incorporating routine cognitive assessments into fall risk evaluation frameworks for cognitively frail individuals, as cognitive deficits are a known contributor to falls.\u003c/p\u003e \u003cp\u003eTaken together, our findings illustrate the multifactorial nature of fall risk in older adults with cognitive frailty, spanning physical, psychological, sensory, and cognitive domains. A prior injurious fall reflects both functional deterioration and psychological consequences (e.g., fear of falling), lower ADL scores signal loss of independence, and elevated anxiety levels highlight the role of affective symptoms in impairing balance and coordination. Likewise, sensory deficits (especially hearing loss) reduce environmental awareness, while central adiposity (large waist circumference) compromises stability. This comprehensive perspective reinforces that elevated fall risk arises from an interplay of diverse factors, underlining the need for multidimensional strategies in risk assessment and prevention.\u003c/p\u003e \u003cp\u003eModel Performance and Methodological Advancements: Beyond identifying risk factors, our machine learning framework achieved outstanding predictive performance. Most classifiers demonstrated near-perfect discrimination (AUC\u0026thinsp;\u0026asymp;\u0026thinsp;1.00) even when restricted to the top 20% of features. Notably, logistic regression achieved comparably high accuracy and AUC while providing greater transparency and interpretability, which is critical for clinical use. This performance clearly exceeds that of conventional fall risk models, which typically report AUCs of 0.65\u0026ndash;0.88 (median\u0026thinsp;~\u0026thinsp;0.72 [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e],); for instance, a recently published model achieved an AUC-ROC of only 0.734 [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Our findings also extend those of Park et al. (2025) [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e], who used logistic regression on longitudinal data from the Korean Frailty and Aging Cohort Study to identify four optimal variables (Fried PF phenotypes, PF-M, SGDS-K, and SARC-F), achieving excellent discrimination (AUC, sensitivity, specificity, and accuracy\u0026thinsp;\u0026ge;\u0026thinsp;91%). While their work demonstrated the feasibility of using a small set of physical and psychological variables, our study incorporated a broader pool of 93 CGA-derived candidate features, from which 58 were retained after rigorous feature selection. We further compared 11 machine learning algorithms under both full- and under-sampled conditions, ensuring that model stability was rigorously assessed across class distributions. Unlike most prior studies that relied on oversampling techniques (e.g., SMOTE), our use of random under-sampling reduced majority-class bias and improved generalizability.\u003c/p\u003e \u003cp\u003eSeveral methodological innovations likely contributed to these excellent results, distinguishing our work from traditional models [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e],. First, we integrated a broad range of multidimensional predictors, combining conventional clinical markers (e.g., known osteoporosis, ADL scores) with psychological, sensory, and lifestyle variables to capture a more holistic profile of fall risk. This inclusive approach goes beyond the narrower, expert-selected feature sets often used in earlier fall risk models. Second, we employed both incremental and decremental feature selection strategies to optimize model parsimony without sacrificing accuracy. Our results demonstrate that high discrimination can be maintained with a substantially reduced feature set, which is critical for real-world feasibility and efficiency in clinical settings. Third, we explicitly addressed class imbalance by applying random under-sampling of the majority class (non-fallers). This procedure improved model generalizability and mitigated bias toward the majority class, an issue often overlooked in previous fall risk studies where imbalanced outcomes can lead to overestimation of model performance. Together, these strategies enhanced our model\u0026rsquo;s robustness, efficiency, and clinical applicability.\u003c/p\u003e \u003cp\u003eStrengths and Limitations: This study has several notable strengths. It is one of the first to focus specifically on fall risk prediction in cognitively frail older adults\u0026mdash;a high-risk subgroup that is often underrepresented in fall prevention research. By leveraging comprehensive CGA data, we were able to integrate a wide array of features across medical, functional, psychological, and sensory domains, which strengthens the ecological validity of our findings. Additionally, we rigorously evaluated multiple modeling techniques and sampling strategies (full-sample vs. under-sampling), which adds confidence in the consistency and reproducibility of the identified predictors and model performance. The use of both complex ensemble models and simpler interpretable models (like logistic regression) highlights the versatility and scalability of our approach for different clinical scenarios.\u003c/p\u003e \u003cp\u003eNevertheless, several limitations must be acknowledged. First, our study was conducted in a single-center geriatric inpatient cohort, which may limit the external generalizability of the findings. The participant population was relatively homogeneous, and model performance should be validated in community-dwelling older adults and more diverse geographic or ethnic populations. Second, the study design was cross-sectional, meaning predictors and fall outcomes were assessed contemporaneously. This limits causal inference\u0026mdash;longitudinal studies are needed to establish temporal relationships and confirm that the identified risk factors indeed precede and predict falls over time. Third, although we observed near-perfect classification performance, the modest sample size and the use of under-sampling techniques raise the possibility of overfitting. Our models may have captured patterns specific to this dataset that do not generalize universally. Therefore, external validation in larger, independent cohorts is essential to verify the stability of the predictor set and the true predictive accuracy of the models. Future research should also explore prospective validation and calibration of these models, as well as the integration of additional relevant features (e.g., gait or balance assessments) that were beyond the scope of our current dataset.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eOverall, this study highlights that machine learning applied to CGA data can provide robust, scalable, and clinically relevant tools for fall risk prediction in cognitively frail older adults. By identifying a reproducible set of core predictors with actionable thresholds, our findings support the development of a specialized fall risk scale and supports the implementation of targeted, multidimensional interventions aimed at reducing falls in this high-risk population.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eContributors\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLihua Chen and Meiwei Zhang were responsible for developing the models and drafting the manuscript. Xintong Liu undertook data collection. Yang L\u0026uuml; and Weihua Yu contributed expertise in clinical study design and revise the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData sharing statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dataset, code, algorithm files, and de-identified results used in this study are not publicly available. However, the data for this study can be shared upon reasonable request to the corresponding author.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no potential competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was approved by the Ethics Committee of The First Affiliated Hospital of Chongqing Medical University (approved on 18 July 2012, No.15) and has been performed in accordance with the ethical standards laid down in the Declaration of Helsinki and its later amendments. The need for written informed consent to participate was waived by the First Affiliated Hospital of Chongqing Medical University ethics committee due to retrospective nature of the study (institutional ethics board approved No: 2012-15)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical Trial Number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eClinical trial number: not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFUNDING\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by grants from Chongqing Talent Plan (cstc2022ycjh-bgzxm0184), Key Project of Technological Innovation and Application Development of Chongqing Science \u0026amp; Technology Bureau (CSTC2021jscx-gksb-N0020), Science Innovation Programs Led by the Academicians in Chongqing under Project (2022YSZX-JSX0002CSTB), Chongqing Medical Key Discipline and Regional Medical Key Discipline Development \u0026nbsp;Project 0201【2022】No. 144 202325 and Program for Youth Innovation in Future Medicine, Chongqing Medical University (W0166).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003e\u003cstrong\u003eOlder Adult Fall Prevention. CDC\u0026rsquo;s injury center uses data and research to save lives. Centers for Disease Control and Prevention website. \u003c/strong\u003e[https://www.cdc.gov/falls/pdf/CDC-DIP_At-a-Glance_Falls_508.pdf]\u003c/li\u003e\n\u003cli\u003eCol\u0026oacute;n-Emeric CS, McDermott CL, Lee DS, Berry SD: \u003cstrong\u003eRisk Assessment and Prevention of Falls in Older Community-Dwelling Adults: A Review\u003c/strong\u003e. \u003cem\u003eJama \u003c/em\u003e2024, \u003cstrong\u003e331\u003c/strong\u003e(16):1397-1406.\u003c/li\u003e\n\u003cli\u003eBurzyńska M, Kopiec T, Pikala M: \u003cstrong\u003eMortality Trends due to Falls in the Group of People in Early (65\u0026ndash;74 Years) and Late (75+) Old Age in Poland in the Years 2000\u0026ndash;2020\u003c/strong\u003e. \u003cem\u003eInternational journal of environmental research and public health \u003c/em\u003e2023, \u003cstrong\u003e20\u003c/strong\u003e(6):5073.\u003c/li\u003e\n\u003cli\u003eNicholson WK, Silverstein M, Wong JB, Barry MJ, Chelmow D, Coker TR, Davis EM, Ja\u0026eacute;n CR, Krousel-Wood M, Lee S\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eInterventions to Prevent Falls in Community-Dwelling Older Adults: US Preventive Services Task Force Recommendation Statement\u003c/strong\u003e. \u003cem\u003eJama \u003c/em\u003e2024, \u003cstrong\u003e332\u003c/strong\u003e(1):51-57.\u003c/li\u003e\n\u003cli\u003eKim KM, Lui L-Y, Cummings SR: \u003cstrong\u003eRecent fall and high imminent risk of fracture in older men and women\u003c/strong\u003e. \u003cem\u003eAge and ageing \u003c/em\u003e2022, \u003cstrong\u003e51\u003c/strong\u003e(6):afac141.\u003c/li\u003e\n\u003cli\u003eWen P, Sun L, Cheng L, Zhou C, Wang Z, Wang Z: \u003cstrong\u003eFall status and risk factors in older Chinese adults: A cross\u003c/strong\u003e\u003cstrong\u003e‐sectional study\u003c/strong\u003e. \u003cem\u003eJournal of clinical nursing \u003c/em\u003e2024, \u003cstrong\u003e33\u003c/strong\u003e(7):2609-2618.\u003c/li\u003e\n\u003cli\u003eKelaiditi E, Cesari M, Canevelli M, Van Kan GA, Ousset P-J, Gillette-Guyonnet S, Ritz P, Duveau F, Soto M, Provencher V: \u003cstrong\u003eCognitive frailty: rational and definition from an (IANA/IAGG) international consensus group\u003c/strong\u003e. \u003cem\u003eThe Journal of nutrition, health and aging \u003c/em\u003e2013, \u003cstrong\u003e17\u003c/strong\u003e(9):726-734.\u003c/li\u003e\n\u003cli\u003eWang S, Wang M, Leung ISH, Ge S, Xu X, Chen S, Yin Y: \u003cstrong\u003eThe association of cognitive frailty and the risk of falls among older adults: A systematic review and meta-analysis\u003c/strong\u003e. \u003cem\u003eInt J Nurs Pract \u003c/em\u003e2023, \u003cstrong\u003e29\u003c/strong\u003e(4):e13181.\u003c/li\u003e\n\u003cli\u003eDemanze Laurence B, Michel L: \u003cstrong\u003eThe fall in older adults: physical and cognitive problems\u003c/strong\u003e. \u003cem\u003eCurrent aging science \u003c/em\u003e2017, \u003cstrong\u003e10\u003c/strong\u003e(3):185-200.\u003c/li\u003e\n\u003cli\u003eHaug CJ, Drazen JM: \u003cstrong\u003eArtificial intelligence and machine learning in clinical medicine, 2023\u003c/strong\u003e. \u003cem\u003eNew England Journal of Medicine \u003c/em\u003e2023, \u003cstrong\u003e388\u003c/strong\u003e(13):1201-1208.\u003c/li\u003e\n\u003cli\u003eLiu C-H, Peng C-H, Huang L-Y, Chen F-Y, Kuo C-H, Wu C-Z, Cheng Y-F: \u003cstrong\u003eComparison of multiple linear regression and machine learning methods in predicting cognitive function in older Chinese type 2 diabetes patients\u003c/strong\u003e. \u003cem\u003eBMC neurology \u003c/em\u003e2024, \u003cstrong\u003e24\u003c/strong\u003e(1):11.\u003c/li\u003e\n\u003cli\u003eLiu C-H, Peng C-H, Huang L-Y, Chen F-Y, Kuo C-H, Wu C-Z, Cheng Y-F: \u003cstrong\u003eThe comparison between multiple linear regression and machine learning methods in predicting cognitive function in Chinese type 2 diabetes\u003c/strong\u003e. 2023.\u003c/li\u003e\n\u003cli\u003eXiaodong C, Lingxiao H, Kewei S, Yafei W, Shaowu L, Ya F: \u003cstrong\u003eInterpretable Machine Learning for Fall Prediction Among Older Adults in China\u003c/strong\u003e. \u003cem\u003eAm J Prev Med \u003c/em\u003e2023, \u003cstrong\u003e65\u003c/strong\u003e(4).\u003c/li\u003e\n\u003cli\u003eAndrews J, Gold L, Reed M, Garcia J, McClelland R, Fitzpatrick A, Hough C, Cawthon P, Covinsky K: \u003cstrong\u003eAppendicular Lean Mass, Grip Strength, and the Development of Hospital-Associated Activities of Daily Living Disability Among Older Adults in the Health ABC Study\u003c/strong\u003e. \u003cem\u003eThe journals of gerontology Series A, Biological sciences and medical sciences \u003c/em\u003e2022, \u003cstrong\u003e77\u003c/strong\u003e(7):1398-1404.\u003c/li\u003e\n\u003cli\u003eQin K, Lin L, Lu C, Chen W, Guo V: \u003cstrong\u003eAssociation between systemic inflammation and activities of daily living disability among Chinese elderly individuals: the mediating role of handgrip strength\u003c/strong\u003e. \u003cem\u003eAging clinical and experimental research \u003c/em\u003e2022, \u003cstrong\u003e34\u003c/strong\u003e(4):767-774.\u003c/li\u003e\n\u003cli\u003eWang J, Li S, Hu Y, Ren L, Yang R, Jiang Y, Yu M, Liu Z, Wu Y, Dong Z\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe moderating role of psychological resilience in the relationship between falls, anxiety and depressive symptoms\u003c/strong\u003e. \u003cem\u003eJ Affect Disord \u003c/em\u003e2023, \u003cstrong\u003e341\u003c/strong\u003e:211-218.\u003c/li\u003e\n\u003cli\u003eMadigan M, Rosenblatt NJ, Grabiner MD: \u003cstrong\u003eObesity as a factor contributing to falls by older adults\u003c/strong\u003e. \u003cem\u003eCurrent obesity reports \u003c/em\u003e2014, \u003cstrong\u003e3\u003c/strong\u003e:348-354.\u003c/li\u003e\n\u003cli\u003eMatter KC, Sinclair SA, Hostetler SG, Xiang H: \u003cstrong\u003eA comparison of the characteristics of injuries between obese and non-obese inpatients\u003c/strong\u003e. \u003cem\u003eObesity (Silver Spring) \u003c/em\u003e2007, \u003cstrong\u003e15\u003c/strong\u003e(10):2384-2390.\u003c/li\u003e\n\u003cli\u003eGopinath B, Rochtchina E, Wang JJ, Schneider J, Leeder SR, Mitchell P: \u003cstrong\u003ePrevalence of age-related hearing loss in older adults: Blue Mountains Study\u003c/strong\u003e. \u003cem\u003eArch Intern Med \u003c/em\u003e2009, \u003cstrong\u003e169\u003c/strong\u003e(4):415-416.\u003c/li\u003e\n\u003cli\u003eSkalska A, Wizner B, Piotrowicz K, Klich-Rączka A, Klimek E, Mossakowska M, Rowiński R, Kozak-Szkopek E, J\u0026oacute;źwiak A, Gąsowski J\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe prevalence of falls and their relation to visual and hearing impairments among a nation-wide cohort of older Poles\u003c/strong\u003e. \u003cem\u003eExp Gerontol \u003c/em\u003e2013, \u003cstrong\u003e48\u003c/strong\u003e(2):140-146.\u003c/li\u003e\n\u003cli\u003eGopinath B, McMahon CM, Burlutsky G, Mitchell P: \u003cstrong\u003eHearing and vision impairment and the 5-year incidence of falls in older adults\u003c/strong\u003e. \u003cem\u003eAge and Ageing \u003c/em\u003e2016, \u003cstrong\u003e45\u003c/strong\u003e(3):409-414.\u003c/li\u003e\n\u003cli\u003eMuir SW, Gopaul K, Montero Odasso MM: \u003cstrong\u003eThe role of cognitive impairment in fall risk among older adults: a systematic review and meta-analysis\u003c/strong\u003e. \u003cem\u003eAge and ageing \u003c/em\u003e2012, \u003cstrong\u003e41\u003c/strong\u003e(3):299-308.\u003c/li\u003e\n\u003cli\u003eDormosh N, van de Loo B, Heymans MW, Schut MC, Medlock S, van Schoor NM, van der Velde N, Abu-Hanna A: \u003cstrong\u003eA systematic review of fall prediction models for community-dwelling older adults: comparison between models based on research cohorts and models based on routinely collected data\u003c/strong\u003e. \u003cem\u003eAge and Ageing \u003c/em\u003e2024, \u003cstrong\u003e53\u003c/strong\u003e(7).\u003c/li\u003e\n\u003cli\u003eChen X, He L, Shi K, Wu Y, Lin S, Fang Y: \u003cstrong\u003eInterpretable Machine Learning for Fall Prediction Among Older Adults in China\u003c/strong\u003e. \u003cem\u003eAm J Prev Med \u003c/em\u003e2023, \u003cstrong\u003e65\u003c/strong\u003e(4):579-586.\u003c/li\u003e\n\u003cli\u003ePark C, Kim N, Kim M, Won CW, Lee BC: \u003cstrong\u003eAdvancing fall risk prediction in older adults with cognitive frailty: A machine learning approach using 2-year clinical data\u003c/strong\u003e. \u003cem\u003ePLoS One \u003c/em\u003e2025, \u003cstrong\u003e20\u003c/strong\u003e(8):e0330672.\u003c/li\u003e\n\u003cli\u003eVan De Loo B, Seppala LJ, Van Der Velde N, Medlock S, Denkinger M, De Groot LC, Kenny R-A, Moriarty F, Rothenbacher D, Stricker B: \u003cstrong\u003eDevelopment of the AD F ICE_IT models for predicting falls and recurrent falls in community-dwelling older adults: pooled analyses of European cohorts with special attention to medication\u003c/strong\u003e. \u003cem\u003eThe Journals of Gerontology: Series A \u003c/em\u003e2022, \u003cstrong\u003e77\u003c/strong\u003e(7):1446-1454.\u003c/li\u003e\n\u003cli\u003eKang L, Chen X, Han P, Ma Y, Jia L, Fu L, Yu H, Wang L, Hou L, Yu X\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA Screening Tool Using Five Risk Factors Was Developed for Fall-Risk Prediction in Chinese Community-Dwelling Elderly Individuals\u003c/strong\u003e. \u003cem\u003eRejuvenation Res \u003c/em\u003e2018, \u003cstrong\u003e21\u003c/strong\u003e(5):416-422.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1.\u0026nbsp;\u003c/strong\u003eComparison of Demographic and Clinical Characteristics Between Fallers and Non-Fallers in Full and Balanced Datasets\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1A.\u0026nbsp;\u003c/strong\u003eComparison of demographic, lifestyle, and health-related characteristics between fallers and non-fallers in the full dataset (n = 814)\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"568\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariables\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 214px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFully Sampled Dataset (n=814)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eNon-fallers (n=648)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003eFallers (n=166)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDemographic Characteristics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eAge (years), mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e76.32 \u0026plusmn; 7.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e78.05 \u0026plusmn; 7.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eGender (Male), n (%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e266 (41.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e61 (36.75%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.313\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eEducation level, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.699\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eIlliteracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e118 (18.21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e29 (17.47%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003ePrimart school\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e157 (24.23%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eJunior school\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e288 (44.44%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e65 (39.16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eUniversity or above\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e85 (13.12%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e32 (19.28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLifestyle and Health Behavior\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eSmoking status, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.028\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eNever smoked\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e488 (75.31%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e141 (84.94%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003ePrevious smoker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e87 (13.43%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e15 (9.04%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eCurrent smoker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e73 (11.27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e10 (6.02%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eDrinking status, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.352\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eNever drank\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e542 (83.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e146 (87.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003ePrevious drinker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e56 (8.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e10 (6.02%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eCurrent drinking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e50 (7.72%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e10 (6.02%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eLifestyle, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eBasic regular\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e473 (72.99%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e114 (68.67%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eHighly regular\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e116 (17.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e34 (20.48%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eIrregular\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e59 (9.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e18 (10.84%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eChronic Diseases and Health Conditions\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eCoronary heart disease, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e226 (34.88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e58 (34.94%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.922\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eHypertension, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e393 (60.65%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e104 (62.65%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.758\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eOsteoporosis, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e137 (21.14%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e52 (31.33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eStroke, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e140 (21.60%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e47 (28.31%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.166\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eDiabetes, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e227 (35.03%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e59 (35.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.619\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eCataract, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e53 (8.18%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e26 (15.66%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eVisual Impairment, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e512 (79.01%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e147 (88.55%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eVision Problems Affecting Daily Activities, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e84 (12.96%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.000001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eHearing Impairment, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e290 (44.75%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e73 (43.98%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.857\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNutritional and Physical Status\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eMNA Screening Result, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.429\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eWell-nourished\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e418 (64.51%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e116 (69.88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eAt risk of malnutrition\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e212 (32.72%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e46 (27.71%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eMalnutrition\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e18 (2.78%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e4 (2.41%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eGastritis or Ulcer, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e131 (20.22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e21 (12.65%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.049\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eWaist Circumference(cm), mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e86.68 \u0026plusmn; 10.44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e86.81 \u0026plusmn; 11.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.686\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eBMI (kg/m\u0026sup2;), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.380\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026lt; 18.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e54 (8.33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e18 (10.84%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e18.5\u0026ndash;24.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e368 (56.79%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e84 (50.60%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e25\u0026ndash;29.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e212 (32.72%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e60 (36.14%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026ge; 30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e14 (2.16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e4 (2.41%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSleep and Mental Health\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003ePSQI Score, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e10.62 \u0026plusmn; 5.29\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e9.46 \u0026plusmn; 5.38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eSleep Duration (hours), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026lt;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e234 (36.11%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e39 (23.49%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e5~6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e144 (22.22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e31 (18.67%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e6~7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e110 (16.98%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e43 (25.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026gt;7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e160 (24.69%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e51 (30.72%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eSleep Latency (minutes), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026le; 15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e220 (33.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e77 (46.39%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e16~30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e131 (20.22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e24 (14.46%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e31~60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e87 (13.43%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e25 (15.06%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026ge; 60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e210 (32.41%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eGDS-5 (depression), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e201 (31.02%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e43 (25.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.199\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eGAD-7 (anxiety), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e221 (34.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e43 (25.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eMMSE Score, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e22.52 \u0026plusmn; 4.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e21.83 \u0026plusmn; 5.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.094\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eADL Score, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e96.27 \u0026plusmn; 4.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e94.55 \u0026plusmn; 5.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.000066\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMulti-Morbidity and Frailty\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eMultiple Chronic Conditions (\u0026ge;2), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e578 (89.20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e154 (92.77%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.172\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eFrailty Status, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003ePre-frailty, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e487 (75.15%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e103 (62.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003eFrailty, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e161 (24.85%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e63 (37.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eLimbs Flutter, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e69 (10.65%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e26 (15.66%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.073\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 172px;\"\u003e\n \u003cp\u003eUrinary Incontinence, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e100 (15.43%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1B.\u0026nbsp;\u003c/strong\u003eComparison of demographic, lifestyle, and health-related characteristics between fallers and non-fallers in the balanced dataset (n = 332)\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"553\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariables\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 236px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eUndersampled Data (n=332)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003eNonfallers (n=166）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003eFallers (n=166）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDemographic Characteristics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eAge (years), mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e76.30 \u0026plusmn; 8.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e78.05 \u0026plusmn; 7.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.028\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eGender (Male), n (%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e76 (45.78%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e61 (36.75%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.094\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eEducation level, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.084\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eIlliteracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e29 (17.47%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e29 (17.47%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003ePrimart school\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e33 (19.88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eJunior school\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e85 (51.20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e65 (39.16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eUniversity or above\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e19 (11.45%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e32 (19.28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eResidence area, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.091\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eUrban\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e113 (68.07%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e126 (75.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eSuburban\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e6 (3.61%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e7 (4.22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eCounty town\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e31 (18.67%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e15 (9.04%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eVillage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e16 (9.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e18 (10.84%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eLiving arrangement, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.549\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eWith spouse or children\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e150 (90.36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e143 (86.14%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eLiving alone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e11 (6.63%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e15 (9.04%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eNursing home\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e5 (3.01%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e7 (4.22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eJob category, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.851\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003ePrimarily physical labor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e88 (53.01%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e86 (51.81%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003ePrimarily mental labor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e66 (39.76%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e70 (42.17%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eOthers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e12 (7.23%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e10 (6.02%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLifestyle and Health Behavior\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eSmoking status, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eNever smoked\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e114 (68.67%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e142 (85.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003ePrevious smoker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e27 (16.27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e14 (8.43%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eCurrent smoker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e25 (15.06%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e10 (6.02%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eDrinking status, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.194\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eNever drank\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e139 (83.73%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e147 (88.55%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003ePrevious drinker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e18 (10.84%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e10 (6.02%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eCurrent drinking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e9 (5.42%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e9 (5.42%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eLifestyle, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eBasic regular\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e126 (75.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e116 (69.88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eHighly regular\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e27 (16.27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e33 (19.88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eIrregular\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e13 (7.83%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e17 (10.24%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eSelf-Rated Health Status, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.167\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003epoor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e96 (57.83%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e90 (54.22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eregular\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e49 (29.52%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e63 (37.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eGood\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e21 (12.65%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e13 (7.83%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eChronic Diseases and Health Conditions\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eCoronary heart disease, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e65 (39.16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e58 (34.94%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.728\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eHypertension, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e98 (59.04%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e104 (62.65%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.506\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eOsteoporosis, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e28 (16.87%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e52 (31.33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eStroke, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e35 (21.08%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e47 (28.31%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.302\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eDiabetes, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e61 (36.75%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e59 (35.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.972\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eCataract, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e24 (14.46%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e26 (15.66%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.723\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eVisual Impairment, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e133 (80.12%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e147 (88.55%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.035\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eVision Problems Affecting Daily Activities, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e27 (16.27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.035\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eHearing Impairment, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e70 (42.17%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e73 (43.98%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.740\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNutritional and Physical Status\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eMNA Screening Result, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.540\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eWell-nourished\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e3 (1.81%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e2 (1.20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eAt risk of malnutrition\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e55 (33.13%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e46 (27.71%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eMalnutrition\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e108 (65.06%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e116 (69.88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eGastritis or Ulcer, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e34 (20.48%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e21 (12.65%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.145\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eWaist Circumference(cm), mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e87.31 \u0026plusmn; 9.72\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e86.81 \u0026plusmn; 11.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.028\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eBMI (kg/m\u0026sup2;), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.452\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026lt; 18.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e2 (1.20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e17 (10.24%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e18.5\u0026ndash;24.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e100 (60.24%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e87 (52.41%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e25\u0026ndash;29.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e52 (31.33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e59 (35.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026ge; 30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e3 (1.81%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e3 (1.81%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSleep and Mental Health\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003ePSQI Score, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e10.45 \u0026plusmn; 5.40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e9.47 \u0026plusmn; 5.35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eSleep Duration (hours), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.058\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026lt;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e55 (33.13%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e41 (24.70%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e5~6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e41 (24.70%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e31 (18.67%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e6~7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e28 (16.87%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e43 (25.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026gt;7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e42 (25.30%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e51 (30.72%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eSleep Latency (minutes), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.030\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026le; 15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e55 (33.13%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e77 (46.39%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e16~30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e33 (19.88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e24 (14.46%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e31~60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e20 (12.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e25 (15.06%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026ge; 60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e58 (34.94%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eGDS-5 (depression), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e58 (34.94%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e43 (25.90%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.074\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eGAD-7 (anxiety), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e57 (34.34%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e32 (19.28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eMMSE Score, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e22.79 \u0026plusmn; 4.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e21.83 \u0026plusmn; 5.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.094\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eADL Score, mean \u0026plusmn; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e96.42 \u0026plusmn; 4.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e94.55 \u0026plusmn; 5.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eChronic Body Pain, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e81 (48.80%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e77 (46.39%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.661\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMulti-Morbidity and Frailty\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eMultiple Chronic Conditions (\u0026ge;2), n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e146 (87.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e154 (92.77%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.137\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eFrailty Status, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.165\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003ePre-frailty, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e115 (69.28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e103 (62.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eFrailty, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e51 (30.72%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e63 (37.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eLimbs Flutter, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e17 (10.24%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e26 (15.66%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.141\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 153px;\"\u003e\n \u003cp\u003eUrinary Incontinence, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 142px;\"\u003e\n \u003cp\u003e25 (15.06%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e40 (24.10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 49px;\"\u003e\n \u003cp\u003e0.038\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eFootnote (for both tables):\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData are presented as mean \u0026plusmn; standard deviation (SD) for continuous variables and n (%) for categorical variables. Between-group differences were assessed using independent-samples t-test for normally distributed continuous variables, Mann\u0026ndash;Whitney U test for non-normally distributed continuous variables, and chi-square test (or Fisher\u0026rsquo;s exact test when appropriate) for categorical variables. P-values \u0026lt; 0.05 were considered statistically significant\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2.\u003c/strong\u003e Comparison of Feature Selection Stability Between Full and Balanced Datasets\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"545\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eFeature\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eTotal Score\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;(Full Dataset)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eTotal Score\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;(Balanced dataset)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eOverlap\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eVision_Act\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eUrine_Incont\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSleep_Time\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSleep_Duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003ePSQI_Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eFried_Stg\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eFH1lnjury\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eCataract\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eADL Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eVision_Def\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eStroke\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSmoke_Stat\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSelf_Hlth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eOsteoporosis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026radic;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNotes：\u003c/p\u003e\n\u003cp\u003e\u0026radic; indicates that the feature was consistently selected in both datasets with identical total scores.\u003c/p\u003e\n\u003cp\u003e\u0026radic;* indicates that the feature was selected in both datasets but with different total scores, suggesting stable yet slightly variable importance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3\u003c/strong\u003e. Performance of Machine Learning Algorithms Across Feature Proportions in Full and Balanced Datasets (Evaluated by AUC-ROC and F1-score) \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"609\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDataset\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFeature Proportion\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 154px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAlgorithm\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC-ROC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 126px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1-score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eFull Datasets\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNearest Neighbors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSupport Vectors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.23\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRandom Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.75\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNaive Bayes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLinear DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.84\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQuadratic DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNeural Net\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.87\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNearest Neighbors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSupport Vectors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.23\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRandom Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNaive Bayes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLinear DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.84\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQuadratic DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNeural Net\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNearest Neighbors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSupport Vectors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.11\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRandom Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.17\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.07\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNaive Bayes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLinear DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQuadratic DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNeural Net\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eBalanced Datasets\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNearest Neighbors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.45\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSupport Vectors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.78\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRandom Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNaive Bayes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLinear DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.79\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQuadratic DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.93\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNeural Net\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.87\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNearest Neighbors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.75\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSupport Vectors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.85\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRandom Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNaive Bayes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLinear DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.83\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQuadratic DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.74\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNeural Net\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.56\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNearest Neighbors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.26\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSupport Vectors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRandom Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.47\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNaive Bayes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.49\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLinear DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.58\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQuadratic DA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNeural Net\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-geriatrics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bgtc","sideBox":"Learn more about [BMC Geriatrics](http://bmcgeriatr.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bgtc/default.aspx","title":"BMC Geriatrics","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Machine learning, Prediction model, Cognitive Frailty, Falls, Older adults","lastPublishedDoi":"10.21203/rs.3.rs-8022235/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8022235/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eFalls are a major cause of disability in older adults, and cognitive frailty confers greater risk than isolated deficits. However, prediction models seldom target this subgroup. This study aimed to develop machine learning (ML)-based fall risk models for cognitively frail older adults using using Comprehensive Geriatric Assessment (CGA) data.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe included 814 hospitalized older adults with cognitive frailty, and corrected class imbalance using random under-sampling (n\u0026thinsp;=\u0026thinsp;332). Eleven machine learning (ML) algorithms were trained using two feature selection strategies (top 100%\u0026ndash;10% vs. bottom 90%\u0026ndash;10%). Feature importance was evaluated through recursive feature elimination (RFE) and model-based approaches, with clinically actionable thresholds also determined.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eSeven key predictors were consistently identified across sampling strategies: First-Ever Fall-Related Injury (FH1-Injury), ADL (Activities of Daily Living) score, Age, Waist Circumference, Hearing Deficit, Generalized Anxiety Disorder-7 (GAD-7) score, and the Mini-Mental State Examination (MMSE) score. Lower ADL (\u0026lt;\u0026thinsp;100) and lower MMSE (\u0026lt;\u0026thinsp;19) scores were associated with increased fall risk, reflecting functional and cognitive decline. Likewise, advanced age (\u0026ge;\u0026thinsp;79 years), higher GAD-7 (\u0026ge;\u0026thinsp;1) scores indicating anxiety symptoms, and greater waist circumference (\u0026ge;\u0026thinsp;90 cm) predicted elevated fall probability. Decision tree, AdaBoost, and gradient boosting achieved near-perfect discrimination (AUC\u0026thinsp;\u0026asymp;\u0026thinsp;1.00), even when limited to the top 20% of features. Logistic regression yielded comparably high accuracy and AUC while maintaining interpretability, making it suitable for clinical deployment.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThis study presents a robust and scalable ML framework that integrates multidimensional CGA data to predict falls in cognitively frail older adults. Our findings support the development of a tailored fall risk scale and inform multidimensional interventions to prevent falls in this vulnerable population.\u003c/p\u003e","manuscriptTitle":"Risk Prediction Models for Falls Among Older Adults Inpatients with Cognitive frailty: Machine Learning Study Based on Comprehensive Geriatric Assessment","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-18 13:57:22","doi":"10.21203/rs.3.rs-8022235/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-02-05T11:46:37+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-29T14:47:19+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-23T08:02:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"273269046756966940599086065275108577632","date":"2025-12-15T03:29:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"130243267319291902479419681544612770134","date":"2025-12-13T01:49:38+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-12-12T19:40:46+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-11-17T05:40:12+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-11-14T14:41:50+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-14T14:39:00+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Geriatrics","date":"2025-11-03T19:35:24+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-geriatrics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bgtc","sideBox":"Learn more about [BMC Geriatrics](http://bmcgeriatr.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bgtc/default.aspx","title":"BMC Geriatrics","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4388d8d7-4fba-45cd-97ab-052d7af3442c","owner":[],"postedDate":"December 18th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-12T13:17:59+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-18 13:57:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8022235","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8022235","identity":"rs-8022235","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}