Interpretable Machine Learning Models for Childhood and Adolescent Obesity Prediction According to Chinese and WHO Standards: A Two-Wave Cross-Sectional Study

Interpretable Machine Learning Models for Childhood and Adolescent Obesity Prediction According to Chinese and WHO Standards: A Two-Wave Cross-Sectional Study | 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 Article Interpretable Machine Learning Models for Childhood and Adolescent Obesity Prediction According to Chinese and WHO Standards: A Two-Wave Cross-Sectional Study Fangjieyi Zheng, Xiaoqian Wang, Mei Xue, Qiong Wang, Wenqian Zhang, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7617689/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objective: This study aimed to develop and validate interpretable machine learning models to predict childhood and adolescent obesity using multi-domain risk factors, and to deploy these models into an accessible online tool for clinical and public health use. Methods: Data were derived from two waves of cross-sectional surveys (2022 and 2024) conducted in Pinggu District, Beijing, involving 22,555 children and adolescents aged 3–18 years. Obesity was defined according to both Chinese and World Health Organization (WHO) standards. Thirty-eight features across five domains (demographic, fetal life, lifestyle, health status, and family factors) were analyzed. Feature selection was performed using least absolute shrinkage and selection operator (LASSO) regression. Four machine learning models—K-nearest neighbors, LightGBM, neural network, and XGBoost—were trained and evaluated using a comprehensive set of 28 performance metrics. Model interpretability was enhanced using SHapley Additive exPlanations (SHAP). The best-performing models were deployed as web-based applications. Results: Five and twelve features were selected for predicting obesity under Chinese and WHO standards, respectively. Age, maternal BMI, paternal BMI, screen time, and birth weight were consistently important across both standards. The neural network model performed best under Chinese standards (AUC = 0.7352), while XGBoost performed best under WHO standards (AUC = 0.7358). SHAP analysis provided global and local interpretations of feature contributions. User-friendly online prediction tools were developed and made publicly available. Conclusion: This study developed interpretable machine learning models that effectively predict childhood and adolescent obesity using a minimal set of clinically relevant features. The deployed tools facilitate individualized risk assessment and may support targeted prevention strategies. Biological sciences/Computational biology and bioinformatics Health sciences/Diseases Health sciences/Health care Health sciences/Medical research Health sciences/Risk factors Machine learning childhood obesity prediction model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 INTRODUCTION Obesity in children and adolescents is currently prevalent, emerging as a global public health challenge and imposing substantial burdens on healthcare systems 1 . Recent meta-analytical evidence indicates that the global prevalence of childhood and adolescent obesity was 8.5%; compared with the 2000–2011 period, its prevalence saw a 1.5-fold increase between 2012 and 2023 2 . Childhood and adolescent obesity rates exhibit substantial heterogeneity across countries, ranging from 0.4% (Vanuatu) to 28.4% (Puerto Rico). In China, the obesity rate of children and adolescents aged 5–19 years was 7.8% for boys and 5.6% for girls in 2016 based on the International Obesity Task Force (IOTF) standards 3 . Childhood and adolescent obesity tends to persist into adulthood 4,5 , causing significant harm to the cardio-metabolic and respiratory systems 6-8 . It is commonly recognized that childhood and adolescent obesity is largely preventable, and prevention—particularly early prevention—through risk factor modification is the most effective approach to addressing this global health challenge. Gaining a thorough understanding of the potential risk profiles underlying the development of childhood and adolescent obesity is critical, as it may constitute a promising target for formulating targeted prevention strategies. China faces a significant clinical challenge: the parallel application of international (WHO) and domestic growth standards, which, in the absence of harmonized criteria, impedes consistent diagnosis of obesity by medical practitioners 9-11 . Internationally, the World Health Organization (WHO) 12 growth references are widely employed, providing a benchmark for global comparisons. Concurrently, the National Health Commission of the People's Republic of China promulgated the most recent weight management guidelines in December 2024 13 . Derived from nationally representative growth standards, these guidelines are intended to more accurately reflect the physical characteristics of local populations, as evidenced by the use of Healthy China children's data.While both standards share the goal of defining overweight and obesity status, they utilize different statistical approaches and reference populations. Consequently, the prevalence estimates and the classification of individuals, particularly those near the diagnostic thresholds, can differ substantially between the WHO standards and the Chinese standards.This discrepancy creates ambiguity in quantifying the true scale of the problem and, critically, may obscure the identification and relative importance of the underlying risk factors when analyzed under different frameworks. A broad range of modifiable factors have been identified to be associated with childhood and/or adolescent obesity. For example, in a population-based cross-sectional study enrolled 4108 school-aged children and adolescents from Pakistan, both short sleep and long sleep showed significant associations with overweight and obesity 14 . Another nationally representative cross-sectional study of 20,584 children and adolescents aged 5–19 years in Ethiopia showed that maternal age, mobile phone ownership, household use of solid fuel, and households having a middle and higher income independently predicted the increased risk of overweight and obesity 15 . The consumption of ultra-processed foods 16 , maternal junk food consumption 17 , high eating speed and late bedtime 18 were additional risk factors in predisposition to childhood obesity. Given the complex nature of childhood obesity, the specific factors involved and their extent of contribution are not fully understood, with multifaceted reasons underlying this. In retrospect, most prior studies have merely analyzed individual factors or factors within a single domain associated with overweight or obesity in children and adolescents 19,20 , while overlooking their additive or synergistic interactions. Indeed, the contribution of any single factor or domain, by itself, may be small and depend on others. Additionally, some studies had limited sample sizes, which limited the power to detect small effects. In current literature, traditional analytical approaches like correlation analysis, linear regression, and Logistic regression have been widely adopted to identify the potential factors attributable to childhood obesity 10 . However, the stringent prerequisites of these approaches such as independence (where observations must not influence one another) and linearity (where relationships between factors are presumed to follow a straight-line pattern) render many analyses unable to meet them, leaving their conclusions untenable or misleading 21,22 . Addressing these gaps requires the comprehensive incorporation of multi-domain factors related to childhood and adolescent obesity, focusing on their complex, nonlinear, and high-dimensional interactions through advanced statistical approaches with a large homogeneous population. Since 2017, our research team has conducted six waves of regional cross-sectional surveys for the growth and development status of preschool-aged and school-aged children and adolescents (ages 3–18 years) in Beijing, gleaning multi-domain data from both children and adolescents and their parents or guardians 23-30 . Based on data from two recent waves, we aimed to explore the association of multi-domain factors with obesity risk in children and adolescents by using machine learning techniques. Specifically, the best-performing machine learning model was selected through hyperparameter tuning and cross-validation; the minimal sufficient feature set was determined using regularization analyses; the best-performing model on the minimal sufficient set was visually inspected using the SHapley Additive exPlanations (SHAP) and deployed into an online obesity prediction tool for translation in clinical and public health contexts. METHODS Study design and participants Two recent waves of our series ongoing cross-sectional surveys were conducted in Pinggu District, Beijing, China: one wave within January–February 2022 and the other wave within February–June 2024. Both waves were implemented using a systematic stratified cluster sampling strategy to recruit children and adolescents 3–18 years of age who attended kindergartens and schools. This strategy ensures representative sampling across strata and clusters of kindergartens or schools. The 2022 wave enrolled participants from 8 primary schools and 18 middle schools; the 2024 wave enrolled participants from 7 kindergartens, 7 primary schools, and 3 middle schools. The conduct of both waves was reviewed and approved by the Ethics Committees of China-Japan Friendship Hospital ( approval No . 2018-93-K67) and Beijing University of Chinese Medicine ( approval No . 2022BZYLL0906), adhering to local laws and institutional protocols. The parents or guardians of all study children and adolescents provided written informed consent prior to participation. To ensure confidentiality, all survey data were anonymized and assigned unique identifiers for tracking purposes. Data collection Data were collected via self-designed electronic questionnaires circulated through the Wenjuanxing platform (https://www.wjx.cn/) and on-site measurements (only weight and height). Before formal circulation, the reliability and validity of questionnaires were assessed through a pre-test on 200 participants, and the internal consistency was high with Cronbach’s alpha coefficients over 0.85. Quality control Prior to data collection, health practitioners and teachers-in-charge from selected kindergartens and schools underwent training on survey procedures and items in questionnaires. They assisted parents or guardians of participating children and adolescents during surveys. In the case of missing data or aberrant values identified, contact was initiated with parents or guardians for certainty. Definition of obesity Body height and weight were measured to nearest 0.1 cm and 0.1 kg, respectively, by health practitioners and teachers-in-charge at kindergartens and schools after receiving standardized training online using pre-mailed measurement equipment to ensure consistency and reduce artificial errors. Body mass index (BMI) is calculated by dividing the weight in kilograms by the square of height in meters. Obesity is defined based on two distinct standards: the WHO standards and the Chinese standards established by the Chinese National Health Commission in December 2024. The WHO standards employ BMI z-scores, which are standardized by age and sex to represent the deviation of a participant’s BMI from the mean value for the specific age and sex group, expressed in standard deviations (SD). For children below five, the 2006 WHO growth standards are applicable; the 2007 WHO growth references are employed for school-aged children and adolescents aged 5–19 years (61–228 months) 12 . According to the standards established by the WHO, obesity is characterized by a BMI z-score that exceeds +2 SD. Similarly, the Chinese standards utilize BMI z-scores, which are standardized by age and sex and expressed in SD units to measure deviation from age- and sex-specific means. For children under 7, the Chinese guidelines delineate obesity as a BMI z-score > +2 SD, aligning with the WHO standards for this age group. For children aged 7–18 years, the Chinese standard defines obesity using specific BMI screening cutoff values established for children and adolescents of the same age and sex group 13 . Feature definition Our questionnaires include items from five domains: (i) demographic factors: date of birth, age, and sex; (ii) foetal and early life factors: gestational week, mode of delivery, pregnancy and delivery order, twin birth, birth length and weight, infancy feeding practices including breastfeeding duration and time to solid food introduction; (iii) lifestyle-related factors: sedentary time, screen time, outdoor activity time, bedtime, eating speed, weekly intake frequencies of sweet foods, night meals, and fast food; (iv) health status: food or drug allergies, number of dental caries, chronic illnesses; (v) family information: family income, parental height and weight, parental education level, and parental reproductive age. The selection of these items was based on prior research and expert knowledge in the field of childhood and adolescent obesity. In line with the machine learning terminology, all study factors derived from questionnaire items including sociodemographic characteristics and health-relevant metrics of study children and adolescents were uniformly termed as features. Statistical analyses Firstly, before implementing machine learning models, data of all features were preprocessed following eight steps: (i) Missingness filtering: Features with >30% missingness were excluded. (ii) Outlier handling: Outliers identified using the interquartile range (IQR) method (Q3+1.5×IQR) were flagged as missing. (iii) Missingness imputation: Five imputed datasets were generated using multiple imputation. The dataset was selected based on the average area under the receiver operating characteristic curve (AUC) across 20 model-dataset combinations (four machine learning models and five imputed datasets). (iv) Correlation assessment: For highly correlated feature pairs (absolute Spearman’s ρ ≥0.8), the feature having lower clinical relevance or weaker association with obesity was removed. (v) Multicollinearity assessment: Features with variance inflation factor (VIF) >5 were removed. (vi) Feature standardization: Features on continuous scales were standardized using Z-score transformation to ensure uniform feature scaling across models. (vii) Feature selection: To mitigate overfitting while retaining meaningful and interpretable features, the least absolute shrinkage and selection operator (LASSO) was used by imposing L1 regularization penalty on regression coefficients. (viii) Dataset splitting: The selected dataset was randomly partitioned into a training subset (N=11,362) and a validation subset (N=7574) at a 6:4 ratio with balanced obesity distributions. Secondly, selected dataset and features were trained and validated by four machine learning models: K-nearest neighbors (KNN), light gradient boosting machine (LightGBM), neural network, and eXtreme gradient boosting (XGBoost). The selection of these models was grounded in prior research within the fields of medicine and machine learning; and they have been widely employed in modeling feature–disease relationships. Specifically, KNN is a non-parametric, instance-based learning algorithm that classifies data points by majority vote of their k nearest neighbors in the feature space. LightGBM enhances memory efficiency and accelerates training through its histogram-based algorithm for building decision trees. Neural networks excel at modeling intricate, non-linear relationships through their architecture of layered, interconnected artificial neurons modeled on biological systems. XGBoost integrates regularization mechanisms to mitigate overfitting without sacrificing computational speed. The hyperparameters of each machine learning model were tuned and optimized via grid search encompassing 30 distinct combinations and five-fold cross-validation 31,32 . Thirdly, machine learning models were appraised from five aspects using 28 metrics: (i) Discrimination: Accuracy, AUC, Balanced Accuracy, Precision, Sensitivity, Specificity, F-beta Score, Matthews Correlation Coefficient (MCC), and Area under the Precision-Recall Curve (PR AUC). (ii) Calibration and error: Binary Brier Score, Classification Error, Logarithmic Loss, and Multiclass Brier Score. (iii) Confusion matrix metrics: True Positives (TP), True Negatives (TN), False Positives (FP), False Negatives (FN), Positive Predictive Value (PPV), Negative Predictive Value (NPV), False Positive Rate (FPR), False Negative Rate (FNR), False Discovery Rate (FDR), False Omission Rate (FOR), True Positive Rate (TPR), and True Negative Rate (TNR). (iv) Diagnostic utility: Diagnostic Odds Ratio. (v) Multiclass performance: Multiclass AUC using One-vs-One (Pairwise and Uniform Average), One-vs-Rest (Pairwise and Uniform Average), and the Multiclass Extension method. Fourthly, machine learning models were interpreted using SHAP, which can enhance model transparency and clinical applicability. SHAP is based on cooperative game theory, and quantifies the precise contribution of individual features to model prediction relying on Shapley values. Models were visually inspected using waterfall plots (illustrate how the values of each feature cumulatively shift the model’s baseline prediction or average risk to arrive at the final obesity risk score for a specific participant) and force plots (deconstruct the prediction into distinct components that either elevate or diminish the calculated obesity risk, offering an intuitive understanding of personalized risk profiles). These visualizations provide crucial clinical insight into the model’s decision logic and aid in communicating individualized risk. Fifthly, to facilitate clinical translation, an interactive, web-based childhood and adolescent obesity prediction was developed and deployed within the R Shiny framework. This tool integrates features selected through LASSO regression and associated with obesity assessed respectively using WHO and Chinese standards. Users securely input their individual data via an interface, processed remotely. The deployed tool delivers two key outputs: personalized obesity risk probability scores generated by the best-performing machine learning model, and participant-specific SHAP waterfall plots visually depicting how much each feature contributes to their predicted obesity risk. This application tool empowers both individuals and clinicians to investigate the effects of specific features on childhood and adolescent obesity susceptibility, and informs personalized prevention approaches. Sixthly, features were compared between the obesity group and the normal weight group using independent t-test, Mann-Whitney U test, or χ² test where appropriate. Continuous features are expressed as mean ± SD if they distribute normally and median (interquartile range, IQR) otherwise. Categorical features are expressed as frequency count and percentage (%). All hypothesis tests were two-sided; P-value<0.001 was deemed statistically significant. All computational and statistical analyses were executed in the R programming environment (version 4.4.3) under RStudio Desktop (2023.12.1 Build 402, Ocean Storm Release). Machine learning workflows were implemented using the mlr3 and mlr3proba packages (version 3). SHAP values and visualizations were generated using the kernelshap and shapviz packages. RESULTS Study pipeline The pipeline for the analytical process of this study is illustrated in Figure 1 . It includes three parts: selection of study participants; feature selection and model construction; model appraisal and deployment. Baseline characteristics Study participants were enrolled from two waves of our ongoing surveys. After strict data cleaning process and data combination, a total of 22,555 children and adolescents aged 3–18 years from Pinggu District, Beijing were initially eligible. Based on the Chinese standards, 3619 children and adolescents were overweight; based on the WHO standards, this number was 4,038. After excluding children and adolescents with overweight, a final sample of 18,936 (normal weight/obesity: 14,721/4215) and 18,517 (normal weight/obesity: 14,730/3,787) children and adolescents were analyzed according to Chinese and WHO criteria, respectively. Their baseline characteristics are presented and compared between obesity and normal weight in Table 1 . Feature selection This study included 37 features, with missingness rates less than 30%. Missing values were handled using multiple imputation; five imputed datasets were generated. To eliminate variation in units and scales, all continuous features from five imputed datasets were standardized using Z-score transformation. Each dataset was randomly partitioned into the training subset and the validation set at a ratio of 6:4, while maintaining balanced obesity prevalence rates. LASSO regression identified five features (age, maternal BMI, paternal BMI, screen time, and birth weight) for childhood and adolescent obesity assessed by the Chinese standards, and 12 features (age, maternal BMI, paternal BMI, screen time, birth weight, father's age at conception, gestational age, outdoor exercise time, complementary feeding time, breastfeeding duration, birth length, and father's weight) by the WHO standards ( Figure 2 ). Model selection Four machine learning models—KNN, LightGBM, neural network, and XGBoost—were trained on the training subset and evaluated on the validation subset of each imputed dataset. The first dataset with average AUC was selected for downstream analyses. Model performance was assessed using a broad panel of metrics for obesity assessed by the Chinese and WHO standards ( Figure 3 and Table 2 ). Specifically, the neural network model demonstrated superior performance in predicting childhood and adolescent obesity based on the Chinese standards, with the highest AUC (0.7352), F-beta score (0.2656), and specificity (0.9697) but the lowest binary Brier score (0.1423). Adopting the WHO standards, the XGBoost model performed best in predicting obesity; it achieved the highest AUC (0.7358), F-beta score (0.2680), and specificity (0.9683) yet the lowest binary Brier score (0.1423). Based on comprehensive and consistent evaluation of multiple metrics, neural network (the Chinese standards) and XGBoost (the WHO standards) were selected as the best-performing models. The optimal hyperparameters of four machine learning models are summarized in Table 3 and Table 4 for the Chinese and WHO standards, respectively. Model interpretation Based on minimal optimal subset of features and best-performing models, SHAP values were used to enhance model interpretability, with higher values indicating stronger contributions of selected features to the prediction of childhood and adolescent obesity assessed separately by the Chinese and WHO standards ( Figure 4 ). For obesity assessed by the Chinese standards, the most important feature was age, followed by maternal BMI, paternal BMI, screen time, and birthweight. For obesity assessed by the WHO criteria, the most important feature was age, followed by maternal BMI, paternal BMI, screen time, birth weight, father's age at conception, gestational age, outdoor exercise time, complementary feeding time, breastfeeding duration, birth length, and father's weight. Bee-swarm plots demonstrated that the selected features exhibited stronger discriminatory ability in distinguishing obesity from normal weight in children and adolescents, as defined by both Chinese and WHO standards. From the local perspective, the individual contribution of selected features in two cases is illustrated by waterfall and force plots in Figure 4 . Both cases predicted to have obesity. For the Chinese standards, obesity risk was conferred by maternal BMI, birthweight, and paternal BMI, while was counterbalanced by screen time and age. For the WHO standards, obesity risk was conferred by maternal BMI, gestation age, birthweight, paternal BMI, breastfeeding duration, birthweight, and father's age at conception, while was counterbalanced by screen time and complementary feeding time. Application tool deployment To facilitate clinical and public health translation, the best-performing models shipping the minimal optimal subset of features were deployed as web-page application tools for childhood and adolescent obesity assessed by the Chinese and WHO standards, respectively ( Figure 5 ). Both tools are convenient and user-friendly, and are publicly available at https://zhengfjy.shinyapps.io/zfjyobesity/ and https://zhengfjy.shinyapps.io/obesityWHO/. DISCUSSION The aim of this two-wave-based cross-sectional study was to use machine learning models and tune hyperparameters to explore multi-domain factors to predict the risk of obesity (vs. normal weight) among 22,555 children and adolescents aged 3–18 years currently living in Beijing, China. To facilitate external comparison, obesity was defined based on both Chinese and WHO standards, and of 37 features analyzed, five and 12 were separately selected through correlation, multicollinearity, and regularization analyses. Two machine learning models were selected that can effectively predict childhood and adolescent obesity based on the Chinese (neural network) and WHO (XGBoost) standards, with descent performance as reflected by 28 metrics. The best-performing models shipping selected features were visually interpreted by global/local SHAP analyses and were deployed to online prediction tools for practical application. This study is the first in medical literature to develop interpretable machine learning models for childhood and adolescent obesity, aiding in formulating targeted strategies for obesity prevention and pediatric burden reduction. Childhood and adolescent obesity is on the rise globally, exhibiting varying prevalence rates—likely attributable to divergent definitional criteria. To enhance the generalizability of our findings, besides the Chinese standards, we also adopted the WHO standards to define obesity. Under both standards, we pinpointed three consistently retained features through LASSO regression—an approach well-suited for high-dimensional data analysis, as it effectively shrinks less relevant features toward zero while retaining the most informative ones. The three shared features include child age, maternal BMI, and paternal BMI, which are widely recognized as robust candidates with strong prediction potential for childhood and adolescent obesity 33-36 . Indeed, obesity is often seen as a contagious epidemic 37 . It is easily understandable that family environment, including parental modeling of eating, can influence children’s dietary behaviors, consolidating the claim by Ethan Cohen-Cole and Jason M. Fletcher 38 that “the spread of obesity is related to the environment in which individuals live. ” In support of this claim, Christaki and Fowler evaluated a densely interconnected social network of 12,067 people assessed repeatedly from 1971 to 2003 as part of the Framingham Heart Study, reporting that network phenomena appeared to be relevant to the biologic and behavioral trait of obesity, and obesity appeared to spread through social ties 39 . Additionally, Ejima and colleagues proposed a mathematical model to quantitatively examine the contribution of genetic and non-genetic effects to assess their influence on obesity prevalence, and found that socially contagious risk factors had a greater overall influence on the distribution of the population with obesity than did spontaneous weight gain risk or mother-to-child obesity transmission risk 40 . Likewise, Huang and colleagues proposed a belief decision model based on the Dempster-Shafer theory to model obesity epidemic as the competing spread of physical inactivity and physical activity, and concluded that social contagion of obesity can be contained and even eradicated through the competing spread of physical activity belief and physical inactivity belief 41 . Above lines of evidence not only underscores the importance of understanding the inter-person dynamics of obesity spreading, but also lends support to the implication of parental obesity in offspring obesity. Differing from prior studies that focused on one aspect or domain relating to childhood obesity, we incorporated data from five domains—demographic factors, foetal and early life factors, lifestyle-related factors, health status, and family information, and these domains were selected based on literature review in the field of obesity. In total, we included 37 features with missingness rate less than 30%—a threshold widely recognized in data-driven research to strike a balance between preserving meaningful variables and ensuring data robustness. Notably, some study features exhibited complex interconnections. For instance, parental BMI and children’s dietary habits showed strong pairwise associations, while household socioeconomic status and physical activity) formed intricate networks of interdependence. Such multicollinearity and interconnectedness are inherent to real-world epidemiological data, reflecting the multifaceted nature of factors influencing childhood obesity. Traditional analytical methods often rely on strict statistical assumptions, including the independence of variables, which are frequently violated in the presence of multicollinearity, leading to unstable estimates or inflated error rates. This limitation underscores the need for more flexible analytical frameworks capable of handling the inherent complexity of our feature set. To shed more light, we adopted four widely used and statistically powerful machine learning models to overcome the multicollinearity and interconnectedness of study features and provide robust estimates and visual inspections of childhood and adolescent obesity risk. It is worth noting two machine learning models were selected after comprehensive hyperparameter tuning and performance comparison that prove the most effective in predicting childhood and adolescent obesity assessed by the Chinese (neural network) and WHO (XGBoost) standards. On one hand, neural network, a cornerstone of modern artificial intelligence, is computational system inspired by the intricate architecture of the human brain, attempting to replicate the brain’s ability to learn patterns, process information, and make decisions by simulating the behavior of interconnected neurons. This model has been widely adopted in the obesity-related literature. For example, Huang and colleagues used deep neural networks to predict obesity status through the analysis of short audio recordings, and demonstrated a global accuracy of 0.70 and a macro F1-score of 0.65 42 . Leo and colleagues adopted lightweight convolutional neural network model for obesity early detection using thermal images, and its accuracy reached 83.08% 43 . In contrast, the accuracy of neural network was 80.68% for childhood and adolescent obesity based on the Chinese standards, exceeding the performance of other three machine learning models. On the other hand, XGBoost is a highly efficient and scalable implementation of the gradient boosting framework, renowned for its superior performance in machine learning competitions and real-world applications. Likewise, XGBoost is popular in the obesogenic field. For example, Wu and colleagues fed top 40 bacterial species to four machine learning models to predict obesity, and found that XGBoost demonstrated the highest predictive accuracy, reaching 81.44% 44 . In the present study, the accuracy of XGBoost in predicting childhood and adolescent obesity was the highest at 80.63%. Besides the selection of best-performing models, we have provided the details of tuning process of optimal hyperparameters for each machine learning model implemented. This process is critical, as there is evidence that the AUC in XGBoost model can be improved by 5% through hypertuning 45 . Specifically, we optimized hyperparameter settings via grid search and cross-validation to enhance computational efficiency and ensure result stability. Based on the selected features and selected machine learning models, we adopted the SHAP method to overcome the “black box” problem of artificial intelligence and explain the results of the childhood and adolescent obesity prediction models. Notably, the discriminatory ability was descent from both global and local perspectives for selected features under the best-performing models. In response to this, our research team translated this analytical framework into a publicly available application tool to predict the probability of childhood and adolescent obesity assessed by both Chinese and WHO standards and meanwhile provide the visual inspection of each feature to this prediction. Doing so facilitates real-world application of our prediction tools for parents, researchers, and clinicians alike. Finally, some limitations should be acknowledged for this study. Firstly, this study is cross-sectional in design, precluding causal inferring. Secondly, all study participants were enrolled from a remote district of Beijing, leaving the representativeness of Beijing an open question. Thirdly, data were collected via electronic questionnaires except weight and height; recall bias cannot be fully excluded. Fourthly, although five domains of items relating to obesity were analyzed, others such as social determinants and environmental exposures are not covered. Despite above study limitations, our large cross-sectional study successfully identified two best-performing interpretable machine learning models for childhood and adolescent obesity assessed by both Chinese and WHO standards. These models demonstrated robust predictive performance. To bridge the gap between research insights and practical utility, our team further operationalized this analytical framework by deploying it into a user-friendly online prediction tool. This study marks a meaningful step toward actionable obesity prediction for children and adolescents, and its true impact will be realized through iterative refinement, cross-context validation, and intentional collaboration across research, clinical, and public health domains. Declarations Acknowledgments: We are grateful to all participating children and their parents or guardians for their positive cooperation, to the kindergarten or school teachers and health practitioners for their generous help, and to all the researchers for their hard work. Author contributions: Z.Z. and W.N. designed the study. F.Z., Q.W and X.W obtained statutory and ethics approvals and contributed to data acquisition. F.Z., M.X., and W.N. performed the statistical analysis. F.Z. and X.W. wrote the first draft. Z.Z and W.N. are the study guarantors. All authors contributed to the article and approved the submitted version. Conflicts of interest: The authors declare no conflict of interest. Funding: This work was supported by the Public Service Development and Reform Pilot Project of Beijing Medical Research Institute (W. Niu), the Capital’s Funds for Health Improvement and Research (Grant Number: 2024-2-1133), and the National Natural Science Foundation of China (Grant Number: 81970042). Ethics statement: This study was approved by the Ethics Committee of china-japan Friendship Hospital (2018-93-K67) and Beijing University of Chinese Medicine (2022BZYLL0906). Written informed consent was provided by the parents or guardians of children enrolled. Data availability statement: The dataset employed in this study contains identifiable minor data and private data therefore it cannot be fully disclosed. For access to the de-identified dataset, please contact us at [email protected] . References Global, regional, and national prevalence of child and adolescent overweight and obesity, 1990-2021, with forecasts to 2050: a forecasting study for the Global Burden of Disease Study 2021. Lancet . Mar 8 2025;405(10481):785-812. doi:10.1016/s0140-6736(25)00397-6 Zhang X, Liu J, Ni Y, et al. Global Prevalence of Overweight and Obesity in Children and Adolescents: A Systematic Review and Meta-Analysis. JAMA Pediatr . Aug 1 2024;178(8):800-813. doi:10.1001/jamapediatrics.2024.1576 Collaboration NCDRF. Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet . Dec 16 2017;390(10113):2627-2642. doi:10.1016/S0140-6736(17)32129-3 Pacheco LS, Blanco E, Burrows R, Reyes M, Lozoff B, Gahagan S. Early Onset Obesity and Risk of Metabolic Syndrome Among Chilean Adolescents. Prev Chronic Dis . Oct 12 2017;14:E93. doi:10.5888/pcd14.170132 Nielsen J, Hulman A, Narayan KMV, Cunningham SA. Body Mass Index Trajectories From Childhood to Adulthood and Age at Onset of Overweight and Obesity: The Influence of Parents' Weight Status. Am J Epidemiol . Oct 20 2022;191(11):1877-1885. doi:10.1093/aje/kwac124 Fan H, Zhu Q, Zhang X. Child Excess Weight Status, Adult Excess Weight Status, and Cardiometabolic Risk Profile. Front Pediatr . 2020;8:301. doi:10.3389/fped.2020.00301 Zhu Z, Yin P. Overweight and obesity: The serious challenge faced by Chinese children and adolescents. J Glob Health . Jul 21 2023;13:03036. doi:10.7189/jogh.13.03036 Ma G, Meyer CL, Jackson-Morris A, et al. The return on investment for the prevention and treatment of childhood and adolescent overweight and obesity in China: a modelling study. Lancet Reg Health West Pac . Feb 2024;43:100977. doi:10.1016/j.lanwpc.2023.100977 Liang Y, Qi Y. Developmental trajectories of adolescent overweight/obesity in China: socio-economic status correlates and health consequences. Public Health . Aug 2020;185:246-253. doi:10.1016/j.puhe.2020.05.013 Li H, Xiang W, Yi Y, et al. Epidemiology of Overweight and Obesity in Early Childhood in China and Associated Factors. Diabetes Metab Syndr Obes . 2025;18:1809-1822. doi:10.2147/dmso.S493135 Wang L, Hui SS. Diagnostic accuracy of different body weight and height-based definitions of childhood obesity in identifying overfat among Chinese children and adolescents: a cross-sectional study. BMC Public Health . Aug 20 2015;15:802. doi:10.1186/s12889-015-2152-0 Organization WH. Obesity and overweight - World Health Organization (WHO). Updated 7 May 2025. https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight China NHCotPsRo. Weight Management Guidelines. https://www.nhc.gov.cn/ylyjs/zcwj/202412/75cb79c171c94def9e768193e65484f7.shtml Tanveer M, Batrakoulis A, Asghar E, et al. Association of sleep duration with overweight and obesity among school-aged children and adolescents in Pakistan-An empirical cross-sectional study. J Educ Health Promot . 2025;14:43. doi:10.4103/jehp.jehp_1453_24 Mussa I, Debella A, Maruta MB, et al. Overweight and obesity trends and association with household wealth index among children aged 5 to 19 years in Ethiopia a multilevel analysis of 2016 EDHS data. Sci Rep . Mar 28 2025;15(1):10788. doi:10.1038/s41598-024-83773-x Al Hourani H, Shhadeh HA, Al-Jawaldeh A. Association between consumption of ultra processed foods and obesity among Jordanian children and adolescents. Sci Rep . Mar 18 2025;15(1):9326. doi:10.1038/s41598-025-93506-3 Dash A, Balla S, Das S, Goli S. Examining the association between maternal junk food consumption and obesity among children: evidence from a cross-sectional survey in India. J Health Popul Nutr . Jun 6 2025;44(1):191. doi:10.1186/s41043-025-00937-9 Liu S, Zhang J, Ma J, et al. Synergistic interaction between bedtime and eating speed in predicting overweight and obesity in Chinese preschool-aged children. Aging (Albany NY) . Apr 12 2019;11(7):2127-2137. doi:10.18632/aging.101906 Liu Y, Trude ACB, Song S, et al. Childhood obesity inequality in northeast China: joint effect of social economic status and school neighborhood environment. BMC Public Health . Feb 13 2023;23(1):318. doi:10.1186/s12889-023-15194-w Li B, Swinburn B. Shifting the way we conceptualise, research and intervene childhood obesity in China and Southeast Asian countries. Lancet Reg Health West Pac . Mar 2025;56:101502. doi:10.1016/j.lanwpc.2025.101502 Fu Y, Gou W, Hu W, et al. Integration of an interpretable machine learning algorithm to identify early life risk factors of childhood obesity among preterm infants: a prospective birth cohort. BMC Med . Jul 10 2020;18(1):184. doi:10.1186/s12916-020-01642-6 LeCroy MN, Kim RS, Stevens J, Hanna DB, Isasi CR. Identifying Key Determinants of Childhood Obesity: A Narrative Review of Machine Learning Studies. Child Obes . Apr 2021;17(3):153-159. doi:10.1089/chi.2020.0324 Chen K, Zheng F, Zhang X, Wang Q, Zhang Z, Niu W. Factors associated with underweight, overweight, and obesity in Chinese children aged 3-14 years using ensemble learning algorithms. J Glob Health . Feb 7 2025;15:04013. doi:10.7189/jogh.15.04013 Zhang X, Wang Q, Gao Z, et al. Prevalence of malnutrition and its associated factors among 18,503 Chinese children aged 3-14 years. Front Nutr . 2023;10:1228799. doi:10.3389/fnut.2023.1228799 Wang Q, Yang M, Chen K, Zheng F, Zhang Z, Niu W. Clustering unhealthy lifestyle factors in Chinese children and adolescents with overweight or obesity. BMC Pediatr . Mar 24 2025;25(1):226. doi:10.1186/s12887-025-05567-y Zheng F, Chen K, Zhang X, Wang Q, Zhang Z, Niu W. Prevalence of malnutrition and associated factors in Chinese children and adolescents aged 3-14 years using machine learning algorithms. J Glob Health . Jul 21 2025;15:04204. doi:10.7189/jogh.15.04204 Pang B, Wang Q, Yang M, et al. Identification and Optimization of Contributing Factors for Precocious Puberty by Machine/Deep Learning Methods in Chinese Girls. Front Endocrinol (Lausanne) . 2022;13:892005. doi:10.3389/fendo.2022.892005 Xue M, Wang Q, Zhang Y, et al. Factors Associated With Lower Respiratory Tract Infection Among Chinese Students Aged 6-14 Years. Front Pediatr . 2022;10:911591. doi:10.3389/fped.2022.911591 Zhou B, Qu X, Li M, et al. Correlation of bone age development with overweight and obesity in 23,305 children from Beijing. Endocrine . Jan 2025;87(1):304-313. doi:10.1007/s12020-024-03988-w Wang Q, Yang M, Deng X, et al. Explorations on risk profiles for overweight and obesity in 9501 preschool-aged children. Obes Res Clin Pract . Mar-Apr 2022;16(2):106-114. doi:10.1016/j.orcp.2022.02.007 Radzi SFM, Karim MKA, Saripan MI, Rahman MAA, Isa INC, Ibahim MJ. Hyperparameter Tuning and Pipeline Optimization via Grid Search Method and Tree-Based AutoML in Breast Cancer Prediction. J Pers Med . Sep 29 2021;11(10)doi:10.3390/jpm11100978 Warkentin MT, Al-Sawaihey H, Lam S, et al. Radiomics analysis to predict pulmonary nodule malignancy using machine learning approaches. Thorax . Mar 15 2024;79(4):307-315. doi:10.1136/thorax-2023-220226 Shaban Mohamed MA, AbouKhatwa MM, Saifullah AA, et al. Risk Factors, Clinical Consequences, Prevention, and Treatment of Childhood Obesity. Children (Basel) . Dec 16 2022;9(12)doi:10.3390/children9121975 Francis LA, Lee Y, Birch LL. Parental weight status and girls' television viewing, snacking, and body mass indexes. Obes Res . Jan 2003;11(1):143-51. doi:10.1038/oby.2003.23 Bambra CL, Hillier FC, Moore HJ, Summerbell CD. Tackling inequalities in obesity: a protocol for a systematic review of the effectiveness of public health interventions at reducing socioeconomic inequalities in obesity amongst children. Syst Rev . Feb 23 2012;1:16. doi:10.1186/2046-4053-1-16 Svensson V, Jacobsson JA, Fredriksson R, et al. Associations between severity of obesity in childhood and adolescence, obesity onset and parental BMI: a longitudinal cohort study. Int J Obes (Lond) . Jan 2011;35(1):46-52. doi:10.1038/ijo.2010.189 Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser . 2000;894:i-xii, 1-253. Cohen-Cole E, Fletcher JM. Is obesity contagious? Social networks vs. environmental factors in the obesity epidemic. J Health Econ . Sep 2008;27(5):1382-7. doi:10.1016/j.jhealeco.2008.04.005 Christakis NA, Fowler JH. The spread of obesity in a large social network over 32 years. N Engl J Med . Jul 26 2007;357(4):370-9. doi:10.1056/NEJMsa066082 Ejima K, Thomas DM, Allison DB. A Mathematical Model for Predicting Obesity Transmission with Both Genetic and Nongenetic Heredity. Obesity (Silver Spring) . May 2018;26(5):927-933. doi:10.1002/oby.22135 Huang H, Yan Z, Chen Y, Liu F. A social contagious model of the obesity epidemic. Sci Rep . Nov 28 2016;6:37961. doi:10.1038/srep37961 Huang J, Guo P, Zhang S, Ji M, An R. Use of Deep Neural Networks to Predict Obesity With Short Audio Recordings: Development and Usability Study. JMIR AI . Jul 25 2024;3:e54885. doi:10.2196/54885 Leo H, Saddami K, Roslidar, Muharar R, Munadi K, Arnia F. Lightweight convolutional neural network (CNN) model for obesity early detection using thermal images. Digit Health . Jan-Dec 2024;10:20552076241271639. doi:10.1177/20552076241271639 Wu H, Li Y, Jiang Y, et al. Machine learning prediction of obesity-associated gut microbiota: identifying Bifidobacterium pseudocatenulatum as a potential therapeutic target. Front Microbiol . 2024;15:1488656. doi:10.3389/fmicb.2024.1488656 Pieszko K, Slomka PJ. Assessing Performance of Machine Learning. JAMA Cardiol . Dec 1 2021;6(12):1465. doi:10.1001/jamacardio.2021.3712 Tables Table 1. Baseline characteristics of children and adolescents with obesity assessed by Chinese and WHO standards and normal weight. Characteristic Chinese standards for obesity definition WHO standards for obesity definition Obesity (N=4215) Normal weight (N=14,721) P Obesity (N=3787) Normal weight (N=14,730) P Demographic factors Age, years 9.33 (6.33, 12.17) 7.25 (4.75, 11.42) <0.001 9.37(6.41,12.17) 6.83(4.75,11.17) <0.001 Boys 2542 (60.31) 7056 (47.93) <0.001 2,596(68.55) 6,838(46.42) <0.001 Weight, kg 52 (35, 70) 23 (17.5,39 ) <0.001 50 (34, 76) 25 (18, 42) <0.001 Height, cm 145 (125, 160) 125 (110, 152) <0.001 156(115, 163) 136 (120, 162) <0.001 Foetal and early life factors Gestational age, weeks 39 (38, 40) 39 (38, 40) 0.001 39(38,40) 39 (38, 40) <0.001 Gravidity 1(1,2) 1 (1,2) <0.001 1(1,2) 1,(1,2) <0.001 Parity 1(1,2) 1(1,2) <0.001 1(1,2) 1(1,2) <0.001 Delivery mode <0.001 <0.001 Vaginal delivery 1,793(42.54) 7,541(51.23) 1,573(41.54) 7,575(51.43) Caesarean section 2,374(56.32) 7,024(47.71) 2,170(57.30) 7,004(47.55) Forceps delivery 48(1.14) 156(1.06) 44(1.16) 151(1.03) Birth length, cm 50(50,52) 50 (50, 52) <0.001 50 (50, 52) 50 (50, 52) <0.001 Birth weight, g 3,450(3,100,3,720) 3,300(3,000,3,600) <0.001 3,495(3,100,3,750) 3,300 (3,000, 3,600) <0.001 Infancy feeding 0.003 <0.001 Pure breastfeeding 2,346(55.66) 8,373(56.88) 2,095(55.32) 8,388(56.95) Partial breastfeeding 1,411(33.48) 5,007(34.01) 1,269(33.51) 5,002(34.09) Non-breastfeeding 458(10.87) 1,341(9.11) 423(11.17) 1,320(8.96) Breastfeeding duration 11(0,15) 12(5,16) <0.001 11(0,15) 12(6,16) <0.001 Time to add solid food 6 (6, 7) 6 (6, 6.5) <0.001 6 (6, 7) 6 (6, 6.5) <0.001 Lifestyle-related factors Screen time (hours per day) 1.29(0.93,2.29) 1.21(0.64,2) <0.001 1.286(1,2.29) 1.143(0.64,2) <0.001 Exercise time (hours per week) per day) 1.29(1,2) 1.29(1,2.29) <0.001 1.286(1,2) 1.29(1,2.29) <0.001 Sleep time (hour per day) 9 (8.29, 9.57) 9.28(8.29,10) <0.001 9(8.29,9.57) 9.29(8.29,10.00) <0.001 Night meals <0.001 <0.001 None or once in a while1–2 times weekly 741(17.58) 4,295(29.18) 583(15.39) 4,410(29.94) 1–3 times weekly 1,1993(28.30) 4,418(30.01) 1,062(28.04) 4,411(29.95) 4–6 times weekly 1,619(38.41) 4,238(28.79) 1,464(38.66) 4,140(28.11) Every day 662(15.71) 1,770(12.02) 678(17.90) 1,769(12.01) Bedtime 9.5(9,10) 10(9,10) <0.001 10(9,10) 9.5(9,10) <0.001 Eating speed (minutes per meal) meal) <0.001 <0.001 <15 min 1,583(37.78) 4,836(32.99) 1,439(38.20) 4,848(33.06) 15–30 min 2,503(59.74) 9,178(62.62) 2,231(59.22) 9,166(62.50) ≥30 min 104(2.48) 643(4.39) 97(2.57) 652(4.45) Fried food intake frequency <0.001 <0.001 None or once in a while1–2 times weekly 554(13.14) 4,326(29.39) 473(12.49) 3,958(26.87) 1–3 times weekly 1,458(34.59) 4047(27.49) 486(12.83) 4,458(30.26) 4–6 times weekly 1,687(40.02) 3,894(26.45) 1,527(40.32) 3,781(25.67) Every day 1,458(34.59) 2,454(16.67) 1,301(34.35) 2,533(17.20) Sweet food intake frequency <0.001 <0.001 None or once in a while1–2 times weekly 867(20.57) 2,195(14.91) 783(20.68) 4,916(33.37) 1–3 times weekly 990(23.49) 6,319(42.93) 891(23.53) 6,277(42.61) 4–6 times weekly 2,094(49.68) 4,837(32.86) 1,877(49.56) 2,154(14.62) Every day 264(6.26) 1,370(9.31) 236(6.23) 1,383(9.39) Health status Upper Respiratory Infections (Past Year) 1(0,2) 1(0,2) 0.03 1(0,2) 1(0,2) 0.011 Lower respiratory infections (past year) 0(0,0) 0(0,0) <0.001 0(0,0) 0(0,0) <0.001 Number of cavities 1,686(40.00) 6,889(46.80) <0.001 1,523(40.22) 6,917(46.96) <0.001 Asthma 273(6.48) 531(3.61) <0.001 256(6.76) 520(3.53) <0.001 Rhinitis 1,194(28.33) 3,287(22.33) <0.001 1,095(28.91) 3,252(22.08) <0.001 Eczema 976(23.16) 3,575(24.29) 0.13 871(23.00) 3,604(24.47) <0.001 Family information Maternal reproductive age, years in years 27.08(24.91,29.84) 27.84(25.42,30.67) <0.001 27 (24.83, 29.83) 27.92(25.50,30.75) <0.001 Paternal reproductive age, years in years 28.17 (26.00, 31.00)) 28 .83(26.50, 32.01) <0.001 28.17(26.00,31.00) 28.84(26.50,32.10) <0.001 Maternal height,cm 161(158,165) 161(159,165) 0.04 161(158,165) 162(159,165) 0.006 Paternal height,cm 174(170,178) 175(170,178) 0.004 174(170,178) 175(170,178) 0.007 Maternal weight,kg 64(57,73) 60(53,65) <0.001 64(57,73.50) 60(53,65) <0.001 Paternal weight,kg 80(74,90) 76(70,85) <0.001 80(75,90) 76(70,85) <0.001 Maternal BMI, kg/m 2 24.34(21.88,27.77) 22.46(20.45,24.92) <0.001 24.31(21.88,27.89) 22.46(20.44,24.89) <0.001 Paternal BMI, kg/m 2 26.89(24.49,30.07) 25.24(23.04,27.76) <0.001 27.02(24.49,30.07) 25.25(23.03,27.76) <0.001 Maternal education attainment <0.001 <0.001 High school degree or below 1,787(42.40) 5,504(37.39) 1,601(42.28) 5,478(37.19) Bachelor’s degree 2,334(55.37) 8,284(56.27) 2,102(55.51) 8,307(56.40) Master’s degree or above 94(2.23) 933(6.34) 84(2.22) 945(6.42) Paternal education attainment <0.001 <0.001 High school degree or below 2,121(50.32) 6,368(43.26) 1,889(49.88) 6,335(43.01) Bachelor’s degree 1,975(46.86) 7,322(49.74) 1,794(47.37) 7,349(49.89) Master’s degree or above 119(2.82) 1,031(7.00) 104(2.75) 1,406(7.10) Family income in CNY per year <0.001 <0.001 <100,000 1,773(44.65) 5,693(40.44) 1,579(44.38) 5,680(40.33) 100,000–300,000 1,795(45.20) 6,073(43.14) 1,620(45.53) 6,074(43.13) ≥300,000 403(10.15) 2,312(16.42) 359(10.09) 2,330(16,54) Family history of diabetes 2,357(55.92) 9,640(65.48) <0.001 2,130(56.25) 9,785(66.43) <0.001 Abbreviations: BMI, body mass index; CNY, Chinese Yuan; WHO, World Health Organization. Data are expressed as median (interquartile range) for continuous factors and count (percentage) for categorical factors, and their between-group comparisons were completed using the Mann-Whitney U test and χ 2 test, respectively. Table 2. Performance comparison of five machine learning models for predicting childhood and adolescent obesity assessed by both Chinese and WHO standards. Metric Chinese standards for obesity definition WHO standards for obesity definition KNN LightGBM Neural network XGBoost KNN LightGBM Neural network XGBoost Chinese standards Accuracy 0.7953 0.7898 0.8068 0.7917 0.7953 0.8056 0.8041 0.8063 Area Under the ROC Curve 0.6708 0.7270 0.7352 0.7298 0.6708 0.7356 0.7312 0.7358 Balanced Accuracy 0.5448 0.5446 0.5705 0.5450 0.5448 0.5579 0.5576 0.5711 Binary Brier Score 0.1548 0.1512 0.1423 0.1504 0.1548 0.1424 0.1431 0.1423 Classification Error 0.2047 0.2102 0.1932 0.2083 0.2047 0.1944 0.1959 0.1937 Diagnostic Odds Ratio 4.1803 6.4686 6.6113 7.6492 4.1803 6.6626 6.1586 6.4374 Fβ-Score 0.1953 0.1833 0.2656 0.1813 0.1953 0.2264 0.2273 0.2680 False Discovery Rate 0.5071 0.3767 0.4085 0.3384 0.5071 0.3992 0.4182 0.4154 False Negatives 995 1113 939 1116 995 975 973 936 False Negative Rate 0.8782 0.8925 0.8288 0.8949 0.8782 0.8605 0.8588 0.8261 False Omission Rate 0.1886 0.2037 0.1796 0.2036 0.1886 0.1842 0.1843 0.1794 False Positives 142 81 134 67 142 105 115 140 False Positive Rate 0.0321 0.0183 0.0303 0.0151 0.0321 0.0237 0.0260 0.0317 Logarithmic Loss 0.6255 0.4691 0.4467 0.4673 0.6255 0.4462 0.4487 0.4465 Multiclass AUC Type 1 Pairwise 0.6708 0.7270 0.7352 0.7298 0.6708 0.7356 0.7312 0.7358 Multiclass AUC Type 1 Unweighted 0.6708 0.7270 0.7352 0.7298 0.6708 0.7356 0.7312 0.7358 Multiclass AUC Type N Pairwise 0.6708 0.7270 0.7352 0.7298 0.6708 0.7356 0.7312 0.7358 Multiclass AUC Type N Unweighted 0.6708 0.7270 0.7352 0.7298 0.6708 0.7356 0.7312 0.7358 Multiclass AUC Macro-Averaged 0.6708 0.7270 0.7352 0.7298 0.6708 0.7356 0.7312 0.7358 Multiclass Brier Score 0.3096 0.3023 0.2847 0.3008 0.3096 0.2848 0.2863 0.2846 Matthews Correlation Coefficient 0.1652 0.1934 0.2409 0.2030 0.1652 0.2195 0.2140 0.2401 Negative Predictive Value 0.8114 0.7963 0.8204 0.7964 0.8114 0.8158 0.8157 0.8206 Positive Predictive Value 0.4929 0.6233 0.5915 0.6616 0.4929 0.6008 0.5818 0.5846 Precision-Recall AUC 0.3481 0.4337 0.4212 0.4445 0.3481 0.4218 0.4170 0.4226 Precision 0.4929 0.6233 0.5915 0.6616 0.4929 0.6008 0.5818 0.5846 Sensitivity 0.1218 0.1075 0.1712 0.1051 0.1218 0.1395 0.1412 0.1739 Specificity 0.9679 0.9817 0.9697 0.9849 0.9679 0.9763 0.9740 0.9683 True Negatives 4280 4352 4288 4366 4280 4317 4307 4282 True Negative Rate 0.9679 0.9817 0.9697 0.9849 0.9679 0.9763 0.9740 0.9683 True Positives 138 134 194 131 138 158 160 197 True Positive Rate 0.1218 0.1075 0.1712 0.1051 0.1218 0.1395 0.1412 0.1739 Abbreviations: AUC, area under the ROC; KNN, K-Nearest Neighbors; LightGBM, light gradient boosting machine; ROC, receiver operating characteristic; XGBoost, eXtreme gradient boosting; WHO, World Health Organization. Table 3. Optimal hyperparameters of five machine learning models for the prediction of childhood and adolescent obesity assessed by Chinese standards. Model Optimal hyperparameters KNN K with 20 LightGBM Learning rate with 0.12; bagging fraction 0.3; max_depth 1; num_threads 1; verbose -1 Neural network MaxNWts 10000; trace FALSE; decay 0.0445; maxit 167; size 5 XGBoost Nrounds 1000; nthread 1; verbose 0; eta 0.01; max_depth 2; subsample 0.8 Abbreviations: KNN, K-nearest neighbor; LightGBM, light gradient boosting machine; XGBoost, eXtreme gradient boosting. Table 4. Optimal hyperparameters of five machine learning models for the prediction of childhood and adolescent obesity assessed by WHO standards. Model Optimal hyperparameters KNN K with 20 LightGBM Learning rate with 0.56; bagging fraction 0.9; max_depth 1; num_threads 1; verbose -1 Neural network MaxNWts 10000; trace FALSE; decay 0.0667; maxit 200; size 3 XGBoost Nrounds 1000; nthread 1; verbose 0; eta 0.01; max_depth 5; subsample 0.5 Abbreviations: KNN, K-nearest neighbor; LightGBM, light gradient boosting machine; XGBoost, eXtreme gradient boosting; WHO, World Health Organization. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-7617689","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":584608455,"identity":"73011c35-a4f8-4b59-a351-6b21d3a1c25f","order_by":0,"name":"Fangjieyi Zheng","email":"","orcid":"","institution":"Capital Center for Children’s health, Capital Medical University, Capital Institute of Pediatrics","correspondingAuthor":false,"prefix":"","firstName":"Fangjieyi","middleName":"","lastName":"Zheng","suffix":""},{"id":584608456,"identity":"f6e8a2cf-82d0-4734-9b68-aab2846558a6","order_by":1,"name":"Xiaoqian Wang","email":"","orcid":"","institution":"Chinese Academy of Medical Sciences \u0026 Peking Union Medical College","correspondingAuthor":false,"prefix":"","firstName":"Xiaoqian","middleName":"","lastName":"Wang","suffix":""},{"id":584608459,"identity":"d6957ad2-b1f2-4180-8bdc-c5013a9648c5","order_by":2,"name":"Mei Xue","email":"","orcid":"","institution":"Beijing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Mei","middleName":"","lastName":"Xue","suffix":""},{"id":584608460,"identity":"1a49c844-7172-457d-9497-b5c44758a6fa","order_by":3,"name":"Qiong Wang","email":"","orcid":"","institution":"Beijing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Qiong","middleName":"","lastName":"Wang","suffix":""},{"id":584608461,"identity":"c7b5a3ed-bfaf-4a5d-9190-37c2f8179f77","order_by":4,"name":"Wenqian Zhang","email":"","orcid":"","institution":"Capital Center for Children’s health, Capital Medical University, Capital Institute of Pediatrics","correspondingAuthor":false,"prefix":"","firstName":"Wenqian","middleName":"","lastName":"Zhang","suffix":""},{"id":584608463,"identity":"c44adc41-7970-4371-a8a4-671b095e995a","order_by":5,"name":"Zhixin Zhang","email":"","orcid":"","institution":"China-Japan Friendship Hospital","correspondingAuthor":false,"prefix":"","firstName":"Zhixin","middleName":"","lastName":"Zhang","suffix":""},{"id":584608464,"identity":"1c9dfc56-e665-40c2-94fd-5326b533c6cc","order_by":6,"name":"Wenquan Niu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3klEQVRIiWNgGAWjYDACZgTrAJSRQLQWNphSQloQgMeAOC3m7DxmEh93WCf2z+759pg3x46Bnz3HgOHnDtxaLJt5zCRnnklPnHHn7HZj3m3JDJI9bwwYe8/g1mJwmMfsNm/b4cSGG7nbpHm3HWAwuJFjwMzYRkDLX6CW+TdynoG12BOlhRGoZcONHDaILRIEtbCV/+xtSzfeeCPN3HDutmQeiTPPCg724tNy/vBmg59t1rLzbiQ/e/B2m50cf3vyxgc/8WiBAnCEsoEIHhBxgKAGZC2jYBSMglEwCjAAAKbyUCoTza0DAAAAAElFTkSuQmCC","orcid":"","institution":"Capital Center for Children’s health, Capital Medical University, Capital Institute of Pediatrics","correspondingAuthor":true,"prefix":"","firstName":"Wenquan","middleName":"","lastName":"Niu","suffix":""}],"badges":[],"createdAt":"2025-09-15 07:53:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7617689/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7617689/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101826880,"identity":"514b1d37-3872-4fd9-bea7-e7bbf38d3d0b","added_by":"auto","created_at":"2026-02-04 05:19:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":166013,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAnalytic pipeline of this cross-sectional study.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u003c/strong\u003e APP, application; KNN, K-Nearest Neighbors Algorithm; LightGBM, light gradient boosting machine; ROC, receiver operating characteristic curve; SHAP, SHapley Additive exPlanations; XGBoost, eXtreme gradient boosting；WHO, World Health Organization .\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7617689/v1/ce37f49131731a4cf3707e47.png"},{"id":101881406,"identity":"b32000fd-78ec-4cae-bf06-bc00b8636ee1","added_by":"auto","created_at":"2026-02-04 15:11:57","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":140541,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLASSO coefficient path diagram screening features and regression cross-validation curves based on the Chinese (Panel A) and WHO standards (Panel B).\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations: \u003c/strong\u003eLASSO, Least Absolute Shrinkage and Selection Operator.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7617689/v1/c7dceddf9cb2099ba3305874.png"},{"id":101826879,"identity":"7eb7fca9-592b-4849-bad4-f86d13ee8eee","added_by":"auto","created_at":"2026-02-04 05:19:23","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":68485,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAUCs (solid lines) of four machine learning models to predict childhood and adolescent obesity assessed by the Chinese (Panel A) and WHO (Panel B) standards.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u003c/strong\u003e AUC, area under the receiver operating characteristic (ROC) curve; LightGBM, light gradient boosting machine; NNET, neural network; XGBoost, eXtreme gradient boosting; WHO, World Health Organization.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7617689/v1/5d75daf1baf8513549b6e545.png"},{"id":101881218,"identity":"02707028-d3cf-4bae-af02-4bcf604d05e2","added_by":"auto","created_at":"2026-02-04 15:10:51","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":147111,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGlobal SHAP interpretation of key features to predict childhood and adolescent obesity assessed by the Chinese (Panel A) and WHO (Panel B) standards in the form of important bar, bee swarm, waterfall and force plots.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u003c/strong\u003e BMI, body mass index; SHAP, SHapley Additive exPlanations .Age_dad, paternal reproductive age; BMI_mom, maternal BMI; BMI_dad, paternal BMI; Screen_time, screen watching time(hours per day); ; Outdoor, outdoor activities (hours per day); Solid_food, time to introduce solid food; Breastfeeding_time, breastfeeding duration; Weight_dad, paternal weight.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7617689/v1/d6297bb34aed9ec6b63ace72.png"},{"id":101826883,"identity":"a4008a6c-1245-46d2-ab3b-4864eeb7bf99","added_by":"auto","created_at":"2026-02-04 05:19:23","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":177863,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOnline application tool for the prediction of childhood and adolescent obesity assessed by Chinese (Panel A) and WHO (Panel B) standards.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7617689/v1/f4924f967347359cea395859.png"},{"id":105898090,"identity":"f9f5bdab-6030-42ee-b59b-08406afc5f3a","added_by":"auto","created_at":"2026-04-01 08:59:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2505713,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7617689/v1/0704f295-7027-442f-b696-ee1f72123764.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Interpretable Machine Learning Models for Childhood and Adolescent Obesity Prediction According to Chinese and WHO Standards: A Two-Wave Cross-Sectional Study","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eObesity in children and adolescents is currently prevalent, emerging as a global public health challenge and imposing substantial burdens on healthcare systems\u0026nbsp;\u003csup\u003e1\u003c/sup\u003e. Recent meta-analytical evidence indicates that the global prevalence of childhood and adolescent obesity was 8.5%; compared with the 2000\u0026ndash;2011 period, its prevalence saw a 1.5-fold increase between 2012 and 2023\u003csup\u003e2\u003c/sup\u003e. Childhood and adolescent obesity rates exhibit substantial heterogeneity across countries, ranging from 0.4% (Vanuatu) to 28.4% (Puerto Rico). In China, the obesity rate of children and adolescents aged 5\u0026ndash;19 years was 7.8% for boys and 5.6% for girls in 2016 based on the International Obesity Task Force (IOTF) standards\u003csup\u003e3\u003c/sup\u003e. Childhood and adolescent obesity tends to persist into adulthood \u003csup\u003e4,5\u003c/sup\u003e, causing significant harm to the cardio-metabolic and respiratory systems \u003csup\u003e6-8\u003c/sup\u003e. It is commonly recognized that childhood and adolescent obesity is largely preventable, and prevention\u0026mdash;particularly early prevention\u0026mdash;through risk factor modification is the most effective approach to addressing this global health challenge. Gaining a thorough understanding of the potential risk profiles underlying the development of childhood and adolescent obesity is critical, as it may constitute a promising target for formulating targeted prevention strategies.\u003c/p\u003e\n\u003cp\u003eChina faces a significant clinical challenge: the parallel application of international (WHO) and domestic growth standards, which, in the absence of harmonized criteria, impedes consistent diagnosis of obesity by medical practitioners\u003csup\u003e9-11\u003c/sup\u003e. Internationally, the World Health Organization (WHO)\u003csup\u003e12\u003c/sup\u003e growth references are widely employed, providing a benchmark for global comparisons. Concurrently, the National Health Commission of the People\u0026apos;s Republic of China promulgated the most recent weight management guidelines in December 2024\u003csup\u003e13\u003c/sup\u003e. Derived from nationally representative growth standards, these guidelines are intended to more accurately reflect the physical characteristics of local populations, as evidenced by the use of Healthy China children\u0026apos;s data.While both standards share the goal of defining overweight and obesity status, they utilize different statistical approaches and reference populations. Consequently, the prevalence estimates and the classification of individuals, particularly those near the diagnostic thresholds, can differ substantially between the WHO standards and the Chinese standards.This discrepancy creates ambiguity in quantifying the true scale of the problem and, critically, may obscure the identification and relative importance of the underlying risk factors when analyzed under different frameworks.\u003c/p\u003e\n\u003cp\u003eA broad range of modifiable factors have been identified to be associated with childhood and/or adolescent obesity. For example, in a population-based cross-sectional study enrolled 4108 school-aged children and adolescents from Pakistan, both short sleep and long sleep showed significant associations with overweight and obesity\u003csup\u003e14\u003c/sup\u003e. Another nationally representative cross-sectional study of 20,584 children and adolescents aged 5\u0026ndash;19 years in Ethiopia showed that maternal age, mobile phone ownership, household use of solid fuel, and households having a middle and higher income independently predicted the increased risk of overweight and obesity\u003csup\u003e15\u003c/sup\u003e . The consumption of ultra-processed foods\u003csup\u003e16\u003c/sup\u003e, maternal junk food consumption\u003csup\u003e17\u003c/sup\u003e, high eating speed and late bedtime\u003csup\u003e18\u003c/sup\u003e\u0026nbsp; were additional risk factors in predisposition to childhood obesity. Given the complex nature of childhood obesity,\u0026nbsp;the specific factors involved and their extent of contribution are not fully understood, with multifaceted reasons underlying this. In retrospect, most prior studies have merely analyzed individual factors or factors within a single domain associated with overweight or obesity in children and adolescents\u0026nbsp;\u003csup\u003e19,20\u003c/sup\u003e, while overlooking their additive or synergistic interactions. Indeed, the contribution of any single factor or domain, by itself, may be small and depend on others. Additionally, some studies had limited sample sizes, which limited the power to detect small effects. In current literature, traditional analytical approaches like correlation analysis, linear regression, and Logistic regression have been widely adopted to identify the potential factors attributable to childhood obesity\u0026nbsp;\u003csup\u003e10\u003c/sup\u003e. However, the stringent prerequisites of these approaches such as independence (where observations must not influence one another) and linearity (where relationships between factors are presumed to follow a straight-line pattern) render many analyses unable to meet them, leaving their conclusions untenable or misleading\u0026nbsp;\u003csup\u003e21,22\u003c/sup\u003e. Addressing these gaps requires the comprehensive incorporation of multi-domain factors related to childhood and adolescent obesity, focusing on their complex, nonlinear, and high-dimensional interactions through advanced statistical approaches with a large homogeneous population.\u003c/p\u003e\n\u003cp\u003eSince 2017, our research team has conducted six waves of regional cross-sectional surveys for the growth and development status of preschool-aged and school-aged children and adolescents (ages 3\u0026ndash;18 years) in Beijing, gleaning multi-domain data from both children and adolescents and their parents or guardians\u003csup\u003e23-30\u003c/sup\u003e. Based on data from two recent waves, we aimed to explore the association of multi-domain factors with obesity risk in children and adolescents by using machine learning techniques. Specifically, the best-performing machine learning model was selected through hyperparameter tuning and cross-validation; the minimal sufficient feature set was determined using regularization analyses; the best-performing model on the minimal sufficient set was visually inspected using the SHapley Additive exPlanations (SHAP) and deployed into an online obesity prediction tool for translation in clinical and public health contexts.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eStudy design and participants\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTwo recent waves of our series ongoing cross-sectional surveys were conducted in Pinggu District, Beijing, China: one wave within January\u0026ndash;February 2022 and the other wave within February\u0026ndash;June 2024. Both waves were implemented using\u0026nbsp;a systematic stratified cluster sampling strategy to recruit children and adolescents 3\u0026ndash;18 years of age who attended kindergartens and schools. This strategy ensures representative sampling across strata and clusters of kindergartens or schools. The 2022 wave enrolled participants from 8 primary schools and 18 middle schools; the 2024 wave enrolled participants from 7 kindergartens, 7 primary schools, and 3 middle schools.\u003c/p\u003e\n\u003cp\u003eThe conduct of both waves was reviewed and approved by the Ethics Committees of China-Japan Friendship Hospital (\u003cem\u003eapproval No\u003c/em\u003e. 2018-93-K67) and Beijing University of Chinese Medicine (\u003cem\u003eapproval No\u003c/em\u003e. 2022BZYLL0906), adhering to local laws and institutional protocols. The parents or guardians of all study children and adolescents provided written informed consent prior to participation. To ensure confidentiality, all survey data were anonymized and assigned unique identifiers for tracking purposes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData collection\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData were collected via self-designed electronic questionnaires circulated through the Wenjuanxing platform (https://www.wjx.cn/) and \u003cem\u003eon-site\u003c/em\u003e measurements (only weight and height). Before formal circulation, the reliability and validity of questionnaires were assessed through a pre-test on 200 participants, and the internal consistency was high with Cronbach\u0026rsquo;s alpha coefficients over 0.85.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eQuality control\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePrior to data collection, health practitioners and teachers-in-charge from selected kindergartens and schools underwent training on survey procedures and items in questionnaires. They assisted parents or guardians of participating children and adolescents during surveys. In the case of missing data or aberrant values identified, contact was initiated with parents or guardians for certainty.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDefinition of obesity\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBody height and weight were measured to nearest 0.1 cm and 0.1 kg, respectively, by health practitioners and teachers-in-charge at kindergartens and schools after receiving standardized training online using pre-mailed measurement equipment to ensure consistency and reduce artificial errors. Body mass index (BMI) is calculated by dividing the weight in kilograms by the square of height in meters.\u003c/p\u003e\n\u003cp\u003eObesity is defined based on two distinct standards: the WHO standards and the Chinese standards established by the Chinese National Health Commission in December 2024.\u003c/p\u003e\n\u003cp\u003eThe WHO standards employ BMI z-scores, which are standardized by age and sex to represent the deviation of a participant\u0026rsquo;s BMI from the mean value for the specific age and sex group, expressed in standard deviations (SD). For children below five, the 2006 WHO growth standards are applicable; the 2007 WHO growth references are employed for school-aged children and adolescents aged 5\u0026ndash;19 years (61\u0026ndash;228 months) \u003csup\u003e12\u003c/sup\u003e. According to the standards established by the WHO, obesity is characterized by a BMI z-score that exceeds +2 SD.\u003c/p\u003e\n\u003cp\u003eSimilarly, the Chinese standards utilize BMI z-scores, which are standardized by age and sex and expressed in SD units to measure deviation from age- and sex-specific means. For children under 7, the Chinese guidelines delineate obesity as a BMI z-score \u0026gt; +2 SD, aligning with the WHO standards for this age group. For children aged 7\u0026ndash;18 years, the Chinese standard defines obesity using specific BMI screening cutoff values established for children and adolescents of the same age and sex group \u003csup\u003e13\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFeature definition\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur questionnaires include items from five domains: (i) demographic factors: date of birth, age, and sex; (ii) foetal and early life factors: gestational week, mode of delivery, pregnancy and delivery order, twin birth, birth length and weight, infancy feeding practices including breastfeeding duration and time to solid food introduction; (iii) lifestyle-related factors: sedentary time, screen time, outdoor activity time, bedtime, eating speed, weekly intake frequencies of sweet foods, night meals, and fast food; (iv) health status: food or drug allergies, number of dental caries, chronic illnesses; (v) family information: family income, parental height and weight, parental education level, and parental reproductive age. The selection of these items was based on prior research and expert knowledge in the field of childhood and adolescent obesity.\u003c/p\u003e\n\u003cp\u003eIn line with the machine learning terminology, all study factors derived from questionnaire items including sociodemographic characteristics and health-relevant metrics of study children and adolescents were uniformly termed as features.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eStatistical analyses\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFirstly, before implementing machine learning models, data of all features were preprocessed following eight steps:\u003c/p\u003e\n\u003cp\u003e(i) Missingness filtering: Features with \u0026gt;30% missingness were excluded.\u003c/p\u003e\n\u003cp\u003e(ii) Outlier handling: Outliers identified using the interquartile range (IQR) method (\u0026lt;Q1\u0026ndash;1.5\u0026times;IQR [interquartile range] or \u0026gt;Q3+1.5\u0026times;IQR) were flagged as missing.\u003c/p\u003e\n\u003cp\u003e(iii) Missingness imputation: Five imputed datasets were generated using multiple imputation. The dataset was selected based on the average area under the receiver operating characteristic curve (AUC) across 20 model-dataset combinations (four machine learning models and five imputed datasets).\u003c/p\u003e\n\u003cp\u003e(iv) Correlation assessment: For highly correlated feature pairs (absolute Spearman\u0026rsquo;s \u0026rho; \u0026ge;0.8), the feature having lower clinical relevance or weaker association with obesity was removed.\u003c/p\u003e\n\u003cp\u003e(v) Multicollinearity assessment: Features with variance inflation factor (VIF) \u0026gt;5 were removed.\u003c/p\u003e\n\u003cp\u003e(vi) Feature standardization: Features on continuous scales were standardized using Z-score transformation to ensure uniform feature scaling across models.\u003c/p\u003e\n\u003cp\u003e(vii) Feature selection: To mitigate overfitting while retaining meaningful and interpretable features, the least absolute shrinkage and selection operator (LASSO) was used by imposing\u0026nbsp;L1 regularization penalty on regression coefficients.\u003c/p\u003e\n\u003cp\u003e(viii) Dataset splitting: The selected dataset was randomly partitioned into a training subset (N=11,362) and a validation subset (N=7574) at a 6:4 ratio with balanced obesity distributions.\u003c/p\u003e\n\u003cp\u003eSecondly, selected dataset and features were trained and validated by four machine learning models: K-nearest neighbors (KNN), light gradient boosting machine (LightGBM), neural network, and eXtreme gradient boosting (XGBoost). The selection of these models was grounded in prior research\u0026nbsp;within the fields of medicine and machine learning; and they have been widely employed in modeling feature\u0026ndash;disease relationships. Specifically, KNN is a non-parametric, instance-based learning algorithm that classifies data points by majority vote of their k nearest neighbors in the feature space. LightGBM enhances memory efficiency and accelerates training through its histogram-based algorithm for building decision trees. Neural networks excel at modeling intricate, non-linear relationships through their architecture of layered, interconnected artificial neurons modeled on biological systems. XGBoost integrates regularization mechanisms to mitigate overfitting without sacrificing computational speed. The hyperparameters of each machine learning model were tuned and optimized via grid search encompassing 30 distinct combinations and five-fold cross-validation \u003csup\u003e31,32\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThirdly, machine learning models were appraised from five aspects using 28 metrics:\u003c/p\u003e\n\u003cp\u003e(i) Discrimination: Accuracy, AUC, Balanced Accuracy, Precision, Sensitivity, Specificity, F-beta Score, Matthews Correlation Coefficient (MCC), and Area under the Precision-Recall Curve (PR AUC).\u003c/p\u003e\n\u003cp\u003e(ii) Calibration and error: Binary Brier Score, Classification Error, Logarithmic Loss, and Multiclass Brier Score.\u003c/p\u003e\n\u003cp\u003e(iii) Confusion matrix metrics: True Positives (TP), True Negatives (TN), False Positives (FP), False Negatives (FN), Positive Predictive Value (PPV), Negative Predictive Value (NPV), False Positive Rate (FPR), False Negative Rate (FNR), False Discovery Rate (FDR), False Omission Rate (FOR), True Positive Rate (TPR), and True Negative Rate (TNR).\u003c/p\u003e\n\u003cp\u003e(iv) Diagnostic utility: Diagnostic Odds Ratio.\u003c/p\u003e\n\u003cp\u003e(v) Multiclass performance: Multiclass AUC using One-vs-One (Pairwise and Uniform Average), One-vs-Rest (Pairwise and Uniform Average), and the Multiclass Extension method.\u003c/p\u003e\n\u003cp\u003eFourthly, machine learning models were interpreted using SHAP, which can enhance model transparency and clinical applicability. SHAP is based on cooperative game theory, and quantifies the precise contribution of individual features to model prediction relying on Shapley values. Models were visually inspected using waterfall plots (illustrate how the values of each feature cumulatively shift the model\u0026rsquo;s baseline prediction or average risk to arrive at the final obesity risk score for a specific participant) and force plots (deconstruct the prediction into distinct components that either elevate or diminish the calculated obesity risk, offering an intuitive understanding of personalized risk profiles). These visualizations provide crucial clinical insight into the model\u0026rsquo;s decision logic and aid in communicating individualized risk.\u003c/p\u003e\n\u003cp\u003eFifthly, to facilitate clinical translation, an interactive, web-based childhood and adolescent obesity prediction was developed and deployed within the R Shiny framework. This tool integrates features selected through LASSO regression and associated with obesity assessed respectively using WHO and Chinese standards. Users securely input their individual data via an interface, processed remotely. The deployed tool delivers two key outputs: personalized obesity risk probability scores generated by the best-performing machine learning model, and participant-specific SHAP waterfall plots visually depicting how much each feature contributes to their predicted obesity risk. This application tool empowers both individuals and clinicians to investigate the effects of specific features on childhood and adolescent obesity susceptibility, and informs personalized prevention approaches.\u003c/p\u003e\n\u003cp\u003eSixthly, features were compared between the obesity group and the normal weight group using independent t-test, Mann-Whitney U test, or \u0026chi;\u0026sup2; test where appropriate. Continuous features are expressed as mean \u0026plusmn; SD if they distribute normally and median (interquartile range, IQR) otherwise. Categorical features are expressed as frequency count and percentage (%). All hypothesis tests were two-sided; P-value\u0026lt;0.001 was deemed statistically significant.\u003c/p\u003e\n\u003cp\u003eAll computational and statistical analyses were executed in the R programming environment (version 4.4.3) under RStudio Desktop (2023.12.1 Build 402, Ocean Storm Release). Machine learning workflows were implemented using the mlr3 and mlr3proba packages (version 3). SHAP values and visualizations were generated using the kernelshap and shapviz packages.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eStudy pipeline\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe pipeline for the analytical process of this study is illustrated in \u003cstrong\u003eFigure 1\u003c/strong\u003e. It includes three parts: selection of study participants; feature selection and model construction; model appraisal and deployment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eBaseline characteristics\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStudy participants were enrolled from two waves of our ongoing surveys. After strict data cleaning process and data combination, a total of 22,555 children and adolescents aged 3\u0026ndash;18 years from Pinggu District, Beijing were initially eligible. Based on the Chinese standards, 3619 children and adolescents were overweight; based on the WHO standards, this number was 4,038. After excluding children and adolescents with overweight, a final sample of 18,936 (normal weight/obesity: 14,721/4215) and 18,517 (normal weight/obesity: 14,730/3,787) children and adolescents were analyzed according to Chinese and WHO criteria, respectively. Their baseline characteristics are presented and compared between obesity and normal weight in \u003cstrong\u003eTable 1\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFeature selection\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study included 37 features, with missingness rates less than 30%. Missing values were handled using multiple imputation; five imputed datasets were generated. To eliminate variation in units and scales,\u0026nbsp;all continuous features from five imputed datasets were standardized using Z-score transformation.\u0026nbsp;Each dataset was randomly partitioned into the training subset and the validation set at a ratio of 6:4, while maintaining balanced obesity prevalence rates. LASSO regression identified five features (age, maternal BMI, paternal BMI, screen time, and birth weight) for childhood and adolescent obesity assessed by the Chinese standards, and 12 features (age, maternal BMI, paternal BMI, screen time, birth weight, father\u0026apos;s age at conception, gestational age, outdoor exercise time, complementary feeding time, breastfeeding duration, birth length, and father\u0026apos;s weight) by the WHO standards (\u003cstrong\u003eFigure 2\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eModel selection\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFour machine learning models\u0026mdash;KNN, LightGBM, neural network, and XGBoost\u0026mdash;were trained on the training subset and evaluated on the validation subset of each imputed dataset. The first dataset with average AUC was selected for downstream analyses. Model performance was assessed using a broad panel of metrics for obesity assessed by the Chinese and WHO standards (\u003cstrong\u003eFigure 3\u003c/strong\u003e and \u003cstrong\u003eTable 2\u003c/strong\u003e). Specifically, the neural network model demonstrated superior performance in predicting childhood and adolescent obesity based on the Chinese standards, with the highest AUC (0.7352), F-beta score (0.2656), and specificity (0.9697) but the lowest binary Brier score (0.1423). Adopting the WHO standards, the XGBoost model performed best in predicting obesity; it achieved the highest AUC (0.7358), F-beta score (0.2680), and specificity (0.9683) yet the lowest binary Brier score (0.1423). Based on comprehensive and consistent evaluation of multiple metrics, neural network (the Chinese standards) and XGBoost (the WHO standards) were selected as the best-performing models. The optimal hyperparameters of four machine learning models are summarized in \u003cstrong\u003eTable 3\u003c/strong\u003e and \u003cstrong\u003eTable 4\u003c/strong\u003e for the Chinese and WHO standards, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eModel interpretation\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBased on minimal optimal subset of features and best-performing models, SHAP values were used to enhance model interpretability, with higher values indicating stronger contributions of selected features to the prediction of childhood and adolescent obesity assessed separately by the Chinese and WHO standards (\u003cstrong\u003eFigure 4\u003c/strong\u003e). For obesity assessed by the Chinese standards, the most important feature was age, followed by maternal BMI, paternal BMI, screen time, and birthweight. For obesity assessed by the WHO criteria, the most important feature was age, followed by maternal BMI, paternal BMI, screen time, birth weight, father\u0026apos;s age at conception, gestational age, outdoor exercise time, complementary feeding time, breastfeeding duration, birth length, and father\u0026apos;s weight. Bee-swarm plots demonstrated that the selected features exhibited stronger discriminatory ability in distinguishing obesity from normal weight in children and adolescents, as defined by both Chinese and WHO standards.\u003c/p\u003e\n\u003cp\u003eFrom the local perspective, the individual contribution of selected features in two cases is illustrated by waterfall and force plots in \u003cstrong\u003eFigure 4\u003c/strong\u003e. Both cases predicted to have obesity. For the Chinese standards, obesity risk was conferred by maternal BMI, birthweight, and paternal BMI, while was counterbalanced by screen time and age. For the WHO standards, obesity risk was conferred by maternal BMI, gestation age, birthweight, paternal BMI, breastfeeding duration, birthweight, and father\u0026apos;s age at conception, while was counterbalanced by screen time and complementary feeding time.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eApplication tool deployment\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo facilitate clinical and public health translation, the best-performing models shipping the minimal optimal subset of features were deployed as web-page application tools for childhood and adolescent obesity assessed by the Chinese and WHO standards, respectively (\u003cstrong\u003eFigure 5\u003c/strong\u003e). Both tools are convenient and user-friendly, and are publicly available at https://zhengfjy.shinyapps.io/zfjyobesity/ and https://zhengfjy.shinyapps.io/obesityWHO/.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThe aim of this two-wave-based cross-sectional study was to use machine learning models and tune hyperparameters to explore multi-domain factors to predict the risk of obesity (vs. normal weight) among 22,555 children and adolescents aged 3\u0026ndash;18 years currently living in Beijing, China. To facilitate external comparison, obesity was defined based on both\u0026nbsp;Chinese and WHO standards, and of 37 features analyzed, five and 12 were separately selected through correlation, multicollinearity, and regularization analyses. Two machine learning models were selected that can effectively predict childhood and adolescent obesity based on the Chinese (neural network) and WHO (XGBoost) standards, with descent performance as reflected by 28 metrics. The best-performing models shipping selected features were visually interpreted by global/local SHAP analyses and were deployed to online prediction tools for practical application. This study is the first in medical literature to develop interpretable machine learning models for childhood and adolescent obesity, aiding in formulating targeted strategies for obesity prevention and pediatric burden reduction.\u003c/p\u003e\n\u003cp\u003eChildhood and adolescent obesity is on the rise globally, exhibiting varying prevalence rates\u0026mdash;likely attributable to divergent definitional criteria. To enhance the generalizability of our findings, besides the Chinese standards, we also adopted the WHO standards to define obesity. Under both standards, we pinpointed three consistently retained features through LASSO regression\u0026mdash;an approach well-suited for high-dimensional data analysis, as it effectively shrinks less relevant features toward zero while retaining the most informative ones. The three shared features include child age, maternal BMI, and paternal BMI, which are widely recognized as robust candidates with strong prediction potential for childhood and adolescent obesity \u003csup\u003e33-36\u003c/sup\u003e. Indeed, obesity is often seen as a contagious epidemic\u003csup\u003e37\u003c/sup\u003e. It is easily understandable that family environment, including parental modeling of eating, can influence children\u0026rsquo;s dietary behaviors, consolidating the claim by Ethan Cohen-Cole and Jason M. Fletcher\u003csup\u003e38\u003c/sup\u003e that \u0026ldquo;the spread of obesity is related to the environment in which individuals live. \u0026rdquo; In support of this claim, Christaki and Fowler evaluated a densely interconnected social network of 12,067 people assessed repeatedly from 1971 to 2003 as part of the Framingham Heart Study, reporting that network phenomena appeared to be relevant to the biologic and behavioral trait of obesity, and obesity appeared to spread through social ties\u003csup\u003e39\u003c/sup\u003e. Additionally, Ejima and colleagues proposed a mathematical model to quantitatively examine the contribution of genetic and non-genetic effects to assess their influence on obesity prevalence, and found that\u0026nbsp;socially contagious risk factors had a greater overall influence on the distribution of the population with obesity than did spontaneous weight gain risk or mother-to-child obesity transmission risk\u003csup\u003e40\u003c/sup\u003e. Likewise, Huang and colleagues\u0026nbsp;proposed a belief decision model based on the Dempster-Shafer theory to model obesity epidemic as the competing spread of physical inactivity and physical activity, and concluded that social contagion of obesity can be contained and even eradicated through the competing spread of physical activity belief and physical inactivity belief\u003csup\u003e41\u003c/sup\u003e. Above lines of evidence not only underscores the importance of understanding the inter-person dynamics of obesity spreading, but also lends support to the implication of parental obesity in offspring obesity.\u003c/p\u003e\n\u003cp\u003eDiffering from prior studies that focused on one aspect or domain relating to childhood obesity, we incorporated data from five domains\u0026mdash;demographic factors, foetal and early life factors, lifestyle-related factors, health status, and family information, and these domains were selected based on literature review in the field of obesity. In total, we included 37 features with missingness rate less than 30%\u0026mdash;a threshold widely recognized in data-driven research to strike a balance between preserving meaningful variables and ensuring data robustness. Notably, some study features exhibited complex interconnections. For instance, parental BMI and children\u0026rsquo;s dietary habits showed strong pairwise associations, while household socioeconomic status and physical activity) formed intricate networks of interdependence. Such multicollinearity and interconnectedness are inherent to real-world epidemiological data, reflecting the multifaceted nature of factors influencing childhood obesity. Traditional analytical methods often rely on strict statistical assumptions, including the independence of variables, which are frequently violated in the presence of multicollinearity, leading to unstable estimates or inflated error rates. This limitation underscores the need for more flexible analytical frameworks capable of handling the inherent complexity of our feature set. To shed more light, we adopted four widely used and statistically powerful machine learning models to overcome the multicollinearity and interconnectedness of study features and provide robust estimates and visual inspections of childhood and adolescent obesity risk.\u003c/p\u003e\n\u003cp\u003eIt is worth noting two machine learning models were selected after comprehensive hyperparameter tuning and performance comparison that prove the most effective in predicting childhood and adolescent obesity assessed by the Chinese (neural network) and WHO (XGBoost) standards. On one hand, neural network, a cornerstone of modern artificial intelligence, is computational system inspired by the intricate architecture of the human brain, attempting to replicate the brain\u0026rsquo;s ability to learn patterns, process information, and make decisions by simulating the behavior of interconnected neurons. This model has been widely adopted in the obesity-related literature. For example, Huang and colleagues used deep neural networks to predict obesity status through the analysis of short audio recordings, and\u0026nbsp;demonstrated a global accuracy of 0.70 and a macro F1-score of 0.65\u003csup\u003e42\u003c/sup\u003e. Leo and colleagues adopted lightweight convolutional neural network model\u0026nbsp;for obesity early detection using thermal images, and its accuracy reached 83.08%\u003csup\u003e43\u003c/sup\u003e. In contrast, the accuracy of neural network was 80.68% for childhood and adolescent obesity based on the Chinese standards, exceeding the performance of other three machine learning models. On the other hand, XGBoost is a highly efficient and scalable implementation of the gradient boosting framework, renowned for its superior performance in machine learning competitions and real-world applications. Likewise, XGBoost is popular in the obesogenic field. For example, Wu and colleagues fed top 40 bacterial species to four machine learning models to predict obesity, and found that XGBoost demonstrated the highest predictive accuracy, reaching 81.44%\u003csup\u003e44\u003c/sup\u003e . In the present study, the accuracy of XGBoost in predicting childhood and adolescent obesity was the highest at 80.63%. Besides the selection of best-performing models, we have provided the details of tuning process of optimal hyperparameters for each machine learning model implemented. This process is critical, as there is evidence that the AUC in XGBoost model can be improved by 5% through hypertuning\u003csup\u003e45\u003c/sup\u003e. Specifically, we optimized hyperparameter settings via grid search and cross-validation to enhance computational efficiency and ensure result stability.\u003c/p\u003e\n\u003cp\u003eBased on the selected features and selected machine learning models, we adopted the SHAP method to overcome the \u0026ldquo;black box\u0026rdquo; problem of artificial intelligence and explain the results of the childhood and adolescent obesity prediction models. Notably, the discriminatory ability was descent from both global and local perspectives for selected features under the best-performing models. In response to this, our research team translated this analytical framework into a publicly available application tool to predict the probability of childhood and adolescent obesity assessed by both Chinese and WHO standards and meanwhile provide the visual inspection of each feature to this prediction. Doing so facilitates real-world application of our prediction tools for parents, researchers, and clinicians alike.\u003c/p\u003e\n\u003cp\u003eFinally, some limitations should be acknowledged for this study. Firstly, this study is cross-sectional in design, precluding causal inferring. Secondly, all study participants were enrolled from a remote district of Beijing, leaving the representativeness of Beijing an open question. Thirdly, data were collected via electronic questionnaires except weight and height; recall bias cannot be fully excluded. Fourthly, although five domains of items relating to obesity were analyzed, others such as social determinants and environmental exposures are not covered.\u003c/p\u003e\n\u003cp\u003eDespite above study limitations, our large cross-sectional study successfully identified two best-performing interpretable machine learning models for childhood and adolescent obesity assessed by both Chinese and WHO standards. These models demonstrated robust predictive performance. To bridge the gap between research insights and practical utility, our team further operationalized this analytical framework by deploying it into a user-friendly online prediction tool. This study marks a meaningful step toward actionable obesity prediction for children and adolescents, and its true impact will be realized through iterative refinement, cross-context validation, and intentional collaboration across research, clinical, and public health domains.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments:\u0026nbsp;\u003c/strong\u003eWe are grateful to all participating children and their parents or guardians for their positive cooperation, to the kindergarten or school teachers and health practitioners for their generous help, and to all the researchers for their hard work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions:\u003c/strong\u003e Z.Z. and W.N. designed the study. F.Z., Q.W and X.W obtained statutory and ethics approvals and contributed to data acquisition. F.Z., M.X., and W.N. performed the statistical analysis. F.Z. and X.W. wrote the first draft. Z.Z and W.N. are the study guarantors. All authors contributed to the article and approved the submitted version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of interest:\u003c/strong\u003e The authors declare no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u0026nbsp;\u003c/strong\u003eThis work was supported by the Public Service Development and Reform Pilot Project of Beijing Medical Research Institute (W. Niu), the Capital\u0026rsquo;s Funds for Health Improvement and Research (Grant Number: 2024-2-1133), and the National Natural Science Foundation of China (Grant Number: 81970042).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics statement:\u0026nbsp;\u003c/strong\u003eThis study was approved by the Ethics Committee of china-japan Friendship Hospital (2018-93-K67) and Beijing University of Chinese Medicine (2022BZYLL0906). Written informed consent was provided by the parents or guardians of children enrolled.\u003c/p\u003e\n\u003cp\u003eData availability statement: The dataset employed in this study contains identifiable minor data and private data therefore it cannot be fully disclosed. For access to the de-identified dataset, please contact us at [email protected].\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eGlobal, regional, and national prevalence of child and adolescent overweight and obesity, 1990-2021, with forecasts to 2050: a forecasting study for the Global Burden of Disease Study 2021. \u003cem\u003eLancet\u003c/em\u003e. Mar 8 2025;405(10481):785-812. doi:10.1016/s0140-6736(25)00397-6\u003c/li\u003e\n\u003cli\u003eZhang X, Liu J, Ni Y, et al. Global Prevalence of Overweight and Obesity in Children and Adolescents: A Systematic Review and Meta-Analysis. \u003cem\u003eJAMA Pediatr\u003c/em\u003e. Aug 1 2024;178(8):800-813. doi:10.1001/jamapediatrics.2024.1576\u003c/li\u003e\n\u003cli\u003eCollaboration NCDRF. Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. \u003cem\u003eLancet\u003c/em\u003e. Dec 16 2017;390(10113):2627-2642. doi:10.1016/S0140-6736(17)32129-3\u003c/li\u003e\n\u003cli\u003ePacheco LS, Blanco E, Burrows R, Reyes M, Lozoff B, Gahagan S. Early Onset Obesity and Risk of Metabolic Syndrome Among Chilean Adolescents. \u003cem\u003ePrev Chronic Dis\u003c/em\u003e. Oct 12 2017;14:E93. doi:10.5888/pcd14.170132\u003c/li\u003e\n\u003cli\u003eNielsen J, Hulman A, Narayan KMV, Cunningham SA. Body Mass Index Trajectories From Childhood to Adulthood and Age at Onset of Overweight and Obesity: The Influence of Parents\u0026apos; Weight Status. \u003cem\u003eAm J Epidemiol\u003c/em\u003e. Oct 20 2022;191(11):1877-1885. doi:10.1093/aje/kwac124\u003c/li\u003e\n\u003cli\u003eFan H, Zhu Q, Zhang X. Child Excess Weight Status, Adult Excess Weight Status, and Cardiometabolic Risk Profile. \u003cem\u003eFront Pediatr\u003c/em\u003e. 2020;8:301. doi:10.3389/fped.2020.00301\u003c/li\u003e\n\u003cli\u003eZhu Z, Yin P. Overweight and obesity: The serious challenge faced by Chinese children and adolescents. \u003cem\u003eJ Glob Health\u003c/em\u003e. Jul 21 2023;13:03036. doi:10.7189/jogh.13.03036\u003c/li\u003e\n\u003cli\u003eMa G, Meyer CL, Jackson-Morris A, et al. The return on investment for the prevention and treatment of childhood and adolescent overweight and obesity in China: a modelling study. \u003cem\u003eLancet Reg Health West Pac\u003c/em\u003e. Feb 2024;43:100977. doi:10.1016/j.lanwpc.2023.100977\u003c/li\u003e\n\u003cli\u003eLiang Y, Qi Y. Developmental trajectories of adolescent overweight/obesity in China: socio-economic status correlates and health consequences. \u003cem\u003ePublic Health\u003c/em\u003e. Aug 2020;185:246-253. doi:10.1016/j.puhe.2020.05.013\u003c/li\u003e\n\u003cli\u003eLi H, Xiang W, Yi Y, et al. Epidemiology of Overweight and Obesity in Early Childhood in China and Associated Factors. \u003cem\u003eDiabetes Metab Syndr Obes\u003c/em\u003e. 2025;18:1809-1822. doi:10.2147/dmso.S493135\u003c/li\u003e\n\u003cli\u003eWang L, Hui SS. Diagnostic accuracy of different body weight and height-based definitions of childhood obesity in identifying overfat among Chinese children and adolescents: a cross-sectional study. \u003cem\u003eBMC Public Health\u003c/em\u003e. Aug 20 2015;15:802. doi:10.1186/s12889-015-2152-0\u003c/li\u003e\n\u003cli\u003eOrganization WH. Obesity and overweight - World Health Organization (WHO). Updated 7 May 2025. https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight\u003c/li\u003e\n\u003cli\u003eChina NHCotPsRo. Weight Management Guidelines. https://www.nhc.gov.cn/ylyjs/zcwj/202412/75cb79c171c94def9e768193e65484f7.shtml\u003c/li\u003e\n\u003cli\u003eTanveer M, Batrakoulis A, Asghar E, et al. Association of sleep duration with overweight and obesity among school-aged children and adolescents in Pakistan-An empirical cross-sectional study. \u003cem\u003eJ Educ Health Promot\u003c/em\u003e. 2025;14:43. doi:10.4103/jehp.jehp_1453_24\u003c/li\u003e\n\u003cli\u003eMussa I, Debella A, Maruta MB, et al. Overweight and obesity trends and association with household wealth index among children aged 5 to 19 years in Ethiopia a multilevel analysis of 2016 EDHS data. \u003cem\u003eSci Rep\u003c/em\u003e. Mar 28 2025;15(1):10788. doi:10.1038/s41598-024-83773-x\u003c/li\u003e\n\u003cli\u003eAl Hourani H, Shhadeh HA, Al-Jawaldeh A. Association between consumption of ultra processed foods and obesity among Jordanian children and adolescents. \u003cem\u003eSci Rep\u003c/em\u003e. Mar 18 2025;15(1):9326. doi:10.1038/s41598-025-93506-3\u003c/li\u003e\n\u003cli\u003eDash A, Balla S, Das S, Goli S. Examining the association between maternal junk food consumption and obesity among children: evidence from a cross-sectional survey in India. \u003cem\u003eJ Health Popul Nutr\u003c/em\u003e. Jun 6 2025;44(1):191. doi:10.1186/s41043-025-00937-9\u003c/li\u003e\n\u003cli\u003eLiu S, Zhang J, Ma J, et al. Synergistic interaction between bedtime and eating speed in predicting overweight and obesity in Chinese preschool-aged children. \u003cem\u003eAging (Albany NY)\u003c/em\u003e. Apr 12 2019;11(7):2127-2137. doi:10.18632/aging.101906\u003c/li\u003e\n\u003cli\u003eLiu Y, Trude ACB, Song S, et al. Childhood obesity inequality in northeast China: joint effect of social economic status and school neighborhood environment. \u003cem\u003eBMC Public Health\u003c/em\u003e. Feb 13 2023;23(1):318. doi:10.1186/s12889-023-15194-w\u003c/li\u003e\n\u003cli\u003eLi B, Swinburn B. Shifting the way we conceptualise, research and intervene childhood obesity in China and Southeast Asian countries. \u003cem\u003eLancet Reg Health West Pac\u003c/em\u003e. Mar 2025;56:101502. doi:10.1016/j.lanwpc.2025.101502\u003c/li\u003e\n\u003cli\u003eFu Y, Gou W, Hu W, et al. Integration of an interpretable machine learning algorithm to identify early life risk factors of childhood obesity among preterm infants: a prospective birth cohort. \u003cem\u003eBMC Med\u003c/em\u003e. Jul 10 2020;18(1):184. doi:10.1186/s12916-020-01642-6\u003c/li\u003e\n\u003cli\u003eLeCroy MN, Kim RS, Stevens J, Hanna DB, Isasi CR. Identifying Key Determinants of Childhood Obesity: A Narrative Review of Machine Learning Studies. \u003cem\u003eChild Obes\u003c/em\u003e. Apr 2021;17(3):153-159. doi:10.1089/chi.2020.0324\u003c/li\u003e\n\u003cli\u003eChen K, Zheng F, Zhang X, Wang Q, Zhang Z, Niu W. Factors associated with underweight, overweight, and obesity in Chinese children aged 3-14 years using ensemble learning algorithms. \u003cem\u003eJ Glob Health\u003c/em\u003e. Feb 7 2025;15:04013. doi:10.7189/jogh.15.04013\u003c/li\u003e\n\u003cli\u003eZhang X, Wang Q, Gao Z, et al. Prevalence of malnutrition and its associated factors among 18,503 Chinese children aged 3-14\u0026thinsp;years. \u003cem\u003eFront Nutr\u003c/em\u003e. 2023;10:1228799. doi:10.3389/fnut.2023.1228799\u003c/li\u003e\n\u003cli\u003eWang Q, Yang M, Chen K, Zheng F, Zhang Z, Niu W. Clustering unhealthy lifestyle factors in Chinese children and adolescents with overweight or obesity. \u003cem\u003eBMC Pediatr\u003c/em\u003e. Mar 24 2025;25(1):226. doi:10.1186/s12887-025-05567-y\u003c/li\u003e\n\u003cli\u003eZheng F, Chen K, Zhang X, Wang Q, Zhang Z, Niu W. Prevalence of malnutrition and associated factors in Chinese children and adolescents aged 3-14 years using machine learning algorithms. \u003cem\u003eJ Glob Health\u003c/em\u003e. Jul 21 2025;15:04204. doi:10.7189/jogh.15.04204\u003c/li\u003e\n\u003cli\u003ePang B, Wang Q, Yang M, et al. Identification and Optimization of Contributing Factors for Precocious Puberty by Machine/Deep Learning Methods in Chinese Girls. \u003cem\u003eFront Endocrinol (Lausanne)\u003c/em\u003e. 2022;13:892005. doi:10.3389/fendo.2022.892005\u003c/li\u003e\n\u003cli\u003eXue M, Wang Q, Zhang Y, et al. Factors Associated With Lower Respiratory Tract Infection Among Chinese Students Aged 6-14 Years. \u003cem\u003eFront Pediatr\u003c/em\u003e. 2022;10:911591. doi:10.3389/fped.2022.911591\u003c/li\u003e\n\u003cli\u003eZhou B, Qu X, Li M, et al. Correlation of bone age development with overweight and obesity in 23,305 children from Beijing. \u003cem\u003eEndocrine\u003c/em\u003e. Jan 2025;87(1):304-313. doi:10.1007/s12020-024-03988-w\u003c/li\u003e\n\u003cli\u003eWang Q, Yang M, Deng X, et al. Explorations on risk profiles for overweight and obesity in 9501 preschool-aged children. \u003cem\u003eObes Res Clin Pract\u003c/em\u003e. Mar-Apr 2022;16(2):106-114. doi:10.1016/j.orcp.2022.02.007\u003c/li\u003e\n\u003cli\u003eRadzi SFM, Karim MKA, Saripan MI, Rahman MAA, Isa INC, Ibahim MJ. Hyperparameter Tuning and Pipeline Optimization via Grid Search Method and Tree-Based AutoML in Breast Cancer Prediction. \u003cem\u003eJ Pers Med\u003c/em\u003e. Sep 29 2021;11(10)doi:10.3390/jpm11100978\u003c/li\u003e\n\u003cli\u003eWarkentin MT, Al-Sawaihey H, Lam S, et al. Radiomics analysis to predict pulmonary nodule malignancy using machine learning approaches. \u003cem\u003eThorax\u003c/em\u003e. Mar 15 2024;79(4):307-315. doi:10.1136/thorax-2023-220226\u003c/li\u003e\n\u003cli\u003eShaban Mohamed MA, AbouKhatwa MM, Saifullah AA, et al. Risk Factors, Clinical Consequences, Prevention, and Treatment of Childhood Obesity. \u003cem\u003eChildren (Basel)\u003c/em\u003e. Dec 16 2022;9(12)doi:10.3390/children9121975\u003c/li\u003e\n\u003cli\u003eFrancis LA, Lee Y, Birch LL. Parental weight status and girls\u0026apos; television viewing, snacking, and body mass indexes. \u003cem\u003eObes Res\u003c/em\u003e. Jan 2003;11(1):143-51. doi:10.1038/oby.2003.23\u003c/li\u003e\n\u003cli\u003eBambra CL, Hillier FC, Moore HJ, Summerbell CD. Tackling inequalities in obesity: a protocol for a systematic review of the effectiveness of public health interventions at reducing socioeconomic inequalities in obesity amongst children. \u003cem\u003eSyst Rev\u003c/em\u003e. Feb 23 2012;1:16. doi:10.1186/2046-4053-1-16\u003c/li\u003e\n\u003cli\u003eSvensson V, Jacobsson JA, Fredriksson R, et al. Associations between severity of obesity in childhood and adolescence, obesity onset and parental BMI: a longitudinal cohort study. \u003cem\u003eInt J Obes (Lond)\u003c/em\u003e. Jan 2011;35(1):46-52. doi:10.1038/ijo.2010.189\u003c/li\u003e\n\u003cli\u003eObesity: preventing and managing the global epidemic. Report of a WHO consultation. \u003cem\u003eWorld Health Organ Tech Rep Ser\u003c/em\u003e. 2000;894:i-xii, 1-253. \u003c/li\u003e\n\u003cli\u003eCohen-Cole E, Fletcher JM. Is obesity contagious? Social networks vs. environmental factors in the obesity epidemic. \u003cem\u003eJ Health Econ\u003c/em\u003e. Sep 2008;27(5):1382-7. doi:10.1016/j.jhealeco.2008.04.005\u003c/li\u003e\n\u003cli\u003eChristakis NA, Fowler JH. The spread of obesity in a large social network over 32 years. \u003cem\u003eN Engl J Med\u003c/em\u003e. Jul 26 2007;357(4):370-9. doi:10.1056/NEJMsa066082\u003c/li\u003e\n\u003cli\u003eEjima K, Thomas DM, Allison DB. A Mathematical Model for Predicting Obesity Transmission with Both Genetic and Nongenetic Heredity. \u003cem\u003eObesity (Silver Spring)\u003c/em\u003e. May 2018;26(5):927-933. doi:10.1002/oby.22135\u003c/li\u003e\n\u003cli\u003eHuang H, Yan Z, Chen Y, Liu F. A social contagious model of the obesity epidemic. \u003cem\u003eSci Rep\u003c/em\u003e. Nov 28 2016;6:37961. doi:10.1038/srep37961\u003c/li\u003e\n\u003cli\u003eHuang J, Guo P, Zhang S, Ji M, An R. Use of Deep Neural Networks to Predict Obesity With Short Audio Recordings: Development and Usability Study. \u003cem\u003eJMIR AI\u003c/em\u003e. Jul 25 2024;3:e54885. doi:10.2196/54885\u003c/li\u003e\n\u003cli\u003eLeo H, Saddami K, Roslidar, Muharar R, Munadi K, Arnia F. Lightweight convolutional neural network (CNN) model for obesity early detection using thermal images. \u003cem\u003eDigit Health\u003c/em\u003e. Jan-Dec 2024;10:20552076241271639. doi:10.1177/20552076241271639\u003c/li\u003e\n\u003cli\u003eWu H, Li Y, Jiang Y, et al. Machine learning prediction of obesity-associated gut microbiota: identifying Bifidobacterium pseudocatenulatum as a potential therapeutic target. \u003cem\u003eFront Microbiol\u003c/em\u003e. 2024;15:1488656. doi:10.3389/fmicb.2024.1488656\u003c/li\u003e\n\u003cli\u003ePieszko K, Slomka PJ. Assessing Performance of Machine Learning. \u003cem\u003eJAMA Cardiol\u003c/em\u003e. Dec 1 2021;6(12):1465. doi:10.1001/jamacardio.2021.3712\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1. Baseline characteristics of children and adolescents with obesity assessed by Chinese and WHO standards and normal weight.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCharacteristic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 36px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eChinese standards\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;for obesity definition\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 36px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWHO standards\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;for obesity definition\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eObesity\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(N=4215)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNormal weight\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(N=14,721)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eObesity\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(N=3787)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNormal weight\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(N=14,730)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eDemographic factors\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eAge, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e9.33 (6.33, 12.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7.25 (4.75, 11.42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e9.37(6.41,12.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6.83(4.75,11.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eBoys\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2542 (60.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7056 (47.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,596(68.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,838(46.42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eWeight, kg\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e52 (35, 70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e23 (17.5,39 )\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e50 (34, 76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e25 (18, 42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eHeight, cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e145 (125, 160)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e125 (110, 152)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e156(115, 163)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e136 (120, 162)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFoetal and early life factors\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eGestational age, weeks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e39 (38, 40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e39 (38, 40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e39(38,40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e39 (38, 40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eGravidity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1 (1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eParity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eDelivery mode\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eVaginal delivery\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,793(42.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7,541(51.23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,573(41.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7,575(51.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eCaesarean section\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,374(56.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7,024(47.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,170(57.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7,004(47.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eForceps delivery\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e48(1.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e156(1.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e44(1.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e151(1.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eBirth length, cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e50(50,52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e50 (50, 52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e50 (50, 52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e50 (50, 52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eBirth weight, g\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e3,450(3,100,3,720)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,300(3,000,3,600)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e3,495(3,100,3,750)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,300 (3,000, 3,600)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eInfancy feeding\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003ePure breastfeeding\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,346(55.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e8,373(56.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,095(55.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e8,388(56.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003ePartial breastfeeding\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,411(33.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5,007(34.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,269(33.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5,002(34.09)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eNon-breastfeeding\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e458(10.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,341(9.11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e423(11.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,320(8.96)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eBreastfeeding duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e11(0,15)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e12(5,16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e11(0,15)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e12(6,16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eTime to add solid food\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e6 (6, 7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6 (6, 6.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e6 (6, 7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6 (6, 6.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eLifestyle-related factors\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eScreen time (hours per day)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1.29(0.93,2.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1.21(0.64,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1.286(1,2.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1.143(0.64,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eExercise time (hours per week)\u003c/p\u003e\n \u003cp\u003eper day)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1.29(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1.29(1,2.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1.286(1,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1.29(1,2.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eSleep time (hour per day)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e9 (8.29, 9.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e9.28(8.29,10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e9(8.29,9.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e9.29(8.29,10.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eNight meals\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eNone or once in a while1\u0026ndash;2 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e741(17.58)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,295(29.18)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e583(15.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,410(29.94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e1\u0026ndash;3 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,1993(28.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,418(30.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,062(28.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,411(29.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e4\u0026ndash;6 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,619(38.41)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,238(28.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,464(38.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,140(28.11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eEvery day\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e662(15.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,770(12.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e678(17.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,769(12.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eBedtime\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e9.5(9,10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e10(9,10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e10(9,10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e9.5(9,10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eEating speed (minutes per meal)\u003c/p\u003e\n \u003cp\u003emeal)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u0026lt;15 min\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,583(37.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,836(32.99)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,439(38.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,848(33.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e15\u0026ndash;30 min\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,503(59.74)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e9,178(62.62)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,231(59.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e9,166(62.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u0026ge;30 min\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e104(2.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e643(4.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e97(2.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e652(4.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eFried food intake frequency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eNone or once in a while1\u0026ndash;2 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e554(13.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,326(29.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e473(12.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,958(26.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e1\u0026ndash;3 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,458(34.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4047(27.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e486(12.83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,458(30.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e4\u0026ndash;6 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,687(40.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,894(26.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,527(40.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,781(25.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eEvery day\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,458(34.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e2,454(16.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,301(34.35)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e2,533(17.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eSweet food intake frequency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eNone or once in a while1\u0026ndash;2 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e867(20.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e2,195(14.91)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e783(20.68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,916(33.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e1\u0026ndash;3 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e990(23.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,319(42.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e891(23.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,277(42.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e4\u0026ndash;6 times weekly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,094(49.68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e4,837(32.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,877(49.56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e2,154(14.62)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eEvery day\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e264(6.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,370(9.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e236(6.23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,383(9.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eHealth status\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eUpper Respiratory Infections (Past Year)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1(0,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1(0,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1(0,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1(0,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eLower respiratory infections (past year)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0(0,0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0(0,0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0(0,0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0(0,0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eNumber of cavities\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,686(40.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,889(46.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,523(40.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,917(46.96)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eAsthma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e273(6.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e531(3.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e256(6.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e520(3.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eRhinitis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,194(28.33)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,287(22.33)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,095(28.91)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,252(22.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eEczema\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e976(23.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,575(24.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e871(23.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e3,604(24.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFamily information\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eMaternal reproductive age, years\u003c/p\u003e\n \u003cp\u003ein years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e27.08(24.91,29.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e27.84(25.42,30.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e27 (24.83, 29.83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e27.92(25.50,30.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003ePaternal reproductive age, years\u003c/p\u003e\n \u003cp\u003ein years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e28.17 (26.00, 31.00))\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e28 .83(26.50, 32.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e28.17(26.00,31.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e28.84(26.50,32.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eMaternal height,cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e161(158,165)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e161(159,165)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e161(158,165)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e162(159,165)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003ePaternal height,cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e174(170,178)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e175(170,178)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e174(170,178)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e175(170,178)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eMaternal weight,kg\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e64(57,73)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e60(53,65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e64(57,73.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e60(53,65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003ePaternal weight,kg\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e80(74,90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e76(70,85)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e80(75,90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e76(70,85)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eMaternal BMI, kg/m\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e24.34(21.88,27.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e22.46(20.45,24.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e24.31(21.88,27.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e22.46(20.44,24.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003ePaternal BMI, kg/m\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e26.89(24.49,30.07)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e25.24(23.04,27.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e27.02(24.49,30.07)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e25.25(23.03,27.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eMaternal education attainment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eHigh school degree or below\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,787(42.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5,504(37.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,601(42.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5,478(37.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eBachelor\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,334(55.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e8,284(56.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,102(55.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e8,307(56.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eMaster\u0026rsquo;s degree or above\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e94(2.23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e933(6.34)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e84(2.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e945(6.42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003ePaternal education attainment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eHigh school degree or below\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,121(50.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,368(43.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,889(49.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,335(43.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eBachelor\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,975(46.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7,322(49.74)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,794(47.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e7,349(49.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eMaster\u0026rsquo;s degree or above\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e119(2.82)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,031(7.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e104(2.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1,406(7.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eFamily income in CNY per year\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u0026lt;100,000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,773(44.65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5,693(40.44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,579(44.38)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5,680(40.33)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e100,000\u0026ndash;300,000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,795(45.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,073(43.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e1,620(45.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e6,074(43.13)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003e\u0026ge;300,000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e403(10.15)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e2,312(16.42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e359(10.09)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e2,330(16,54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 26px;\"\u003e\n \u003cp\u003eFamily history of diabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,357(55.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e9,640(65.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e2,130(56.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e9,785(66.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u003c/strong\u003e BMI, body mass index; CNY, Chinese Yuan; WHO, World Health Organization.\u003c/p\u003e\n\u003cp\u003eData are expressed as median (interquartile range) for continuous factors and count (percentage) for categorical factors, and their between-group comparisons were completed using the Mann-Whitney U test and \u0026chi;\u003csup\u003e2\u003c/sup\u003e test, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2. Performance comparison of five machine learning models for predicting childhood and adolescent obesity assessed by both Chinese and WHO standards.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 23px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMetric\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 37px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eChinese standards\u0026nbsp;for obesity definition\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 37px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWHO standards for obesity definition\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKNN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLightGBM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNeural network\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKNN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLightGBM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNeural network\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eChinese standards\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.7953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7898\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.8068\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7917\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.7953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.8056\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.8041\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.8063\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eArea Under the ROC Curve\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7270\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.7352\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7312\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7358\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eBalanced Accuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.5448\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.5446\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.5705\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.5450\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.5448\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.5579\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.5576\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.5711\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eBinary Brier Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.1548\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1512\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.1423\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1504\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.1548\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1424\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1431\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1423\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eClassification Error\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.2047\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2102\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.1932\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2083\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.2047\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1944\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1959\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1937\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eDiagnostic Odds Ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e4.1803\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e6.4686\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e6.6113\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e7.6492\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e4.1803\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e6.6626\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e6.1586\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e6.4374\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eF\u0026beta;-Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.1953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1833\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.2656\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1813\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.1953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2264\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.2273\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.2680\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eFalse Discovery Rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.5071\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.3767\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.4085\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.3384\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.5071\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.3992\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.4182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.4154\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eFalse Negatives\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e995\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e1113\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e939\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e1116\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e995\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e975\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e973\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e936\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eFalse Negative Rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.8782\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.8925\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.8288\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.8949\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.8782\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.8605\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.8588\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.8261\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eFalse Omission Rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.1886\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2037\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.1796\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2036\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.1886\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1842\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1843\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1794\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eFalse Positives\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e134\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e105\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e115\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e140\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eFalse Positive Rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.0321\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.0183\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.0303\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.0151\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.0321\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.0237\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.0260\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.0317\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eLogarithmic Loss\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.6255\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.4691\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.4467\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.4673\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.6255\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.4462\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.4487\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.4465\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eMulticlass AUC Type 1 Pairwise\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7270\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.7352\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7312\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7358\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eMulticlass AUC Type 1 Unweighted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7270\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.7352\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7312\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7358\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eMulticlass AUC Type N Pairwise\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7270\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.7352\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7312\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7358\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eMulticlass AUC Type N Unweighted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7270\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.7352\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7312\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7358\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eMulticlass AUC Macro-Averaged\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7270\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.7352\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.6708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7312\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.7358\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eMulticlass Brier Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.3096\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.3023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.2847\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.3008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.3096\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2848\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.2863\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.2846\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eMatthews Correlation Coefficient\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.1652\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1934\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.2409\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2030\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.1652\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.2195\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.2140\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.2401\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eNegative Predictive Value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.8114\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7963\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.8204\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.7964\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.8114\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.8158\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.8157\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.8206\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003ePositive Predictive Value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.4929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.6233\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.5915\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.6616\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.4929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.6008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.5818\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.5846\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003ePrecision-Recall AUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.3481\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.4337\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.4212\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.4445\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.3481\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.4218\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.4170\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.4226\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.4929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.6233\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.5915\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.6616\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.4929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.6008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.5818\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.5846\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.1218\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1075\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.1712\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1051\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.1218\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1395\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1412\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1739\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.9679\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.9817\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.9697\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.9849\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.9679\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.9763\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.9740\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.9683\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eTrue Negatives\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e4280\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e4352\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4288\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e4366\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e4280\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e4317\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e4307\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e4282\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eTrue Negative Rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.9679\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.9817\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.9697\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.9849\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.9679\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.9763\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.9740\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.9683\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eTrue Positives\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e138\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e134\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e194\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e131\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e138\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e158\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e160\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e197\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 23px;\"\u003e\n \u003cp\u003eTrue Positive Rate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 7px;\"\u003e\n \u003cp\u003e0.1218\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1075\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.1712\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1051\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 1px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.1218\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e0.1395\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1412\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11px;\"\u003e\n \u003cp\u003e0.1739\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u0026nbsp;\u003c/strong\u003eAUC, area under the ROC; KNN, K-Nearest Neighbors; LightGBM, light gradient boosting machine; ROC, receiver operating characteristic; XGBoost, eXtreme gradient boosting; WHO, World Health Organization.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3. Optimal hyperparameters of five machine learning models for the prediction of childhood and adolescent obesity assessed by Chinese standards.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOptimal\u0026nbsp;\u003c/strong\u003ehyperparameters\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eKNN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003eK with 20\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eLightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003eLearning rate with 0.12; bagging fraction 0.3; max_depth 1; num_threads 1; verbose -1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eNeural network\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003eMaxNWts 10000; trace FALSE; decay 0.0445; maxit 167; size 5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003eNrounds 1000; nthread 1; verbose 0; eta 0.01; max_depth 2; subsample 0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAbbreviations: KNN, K-nearest neighbor; LightGBM, light gradient boosting machine; XGBoost, eXtreme gradient boosting.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4. Optimal hyperparameters of five machine learning models for the prediction of childhood and adolescent obesity assessed by WHO standards.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"631\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 175px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 455px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOptimal hyperparameters\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eKNN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eK with 20\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eLightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLearning rate with 0.56; bagging fraction 0.9; max_depth 1; num_threads 1; verbose -1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eNeural network\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eMaxNWts 10000; trace FALSE; decay 0.0667; maxit 200; size 3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNrounds 1000; nthread 1; verbose 0; eta 0.01; max_depth 5; subsample 0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAbbreviations: KNN, K-nearest neighbor; LightGBM, light gradient boosting machine; XGBoost, eXtreme gradient boosting; WHO, World Health Organization.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Machine learning, childhood obesity, prediction model","lastPublishedDoi":"10.21203/rs.3.rs-7617689/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7617689/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective: \u003c/strong\u003eThis study aimed to develop and validate interpretable machine learning models to predict childhood and adolescent obesity using multi-domain risk factors, and to deploy these models into an accessible online tool for clinical and public health use.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eData were derived from two waves of cross-sectional surveys (2022 and 2024) conducted in Pinggu District, Beijing, involving 22,555 children and adolescents aged 3–18 years. Obesity was defined according to both Chinese and World Health Organization (WHO) standards. Thirty-eight features across five domains (demographic, fetal life, lifestyle, health status, and family factors) were analyzed. Feature selection was performed using least absolute shrinkage and selection operator (LASSO) regression. Four machine learning models—K-nearest neighbors, LightGBM, neural network, and XGBoost—were trained and evaluated using a comprehensive set of 28 performance metrics. Model interpretability was enhanced using SHapley Additive exPlanations (SHAP). The best-performing models were deployed as web-based applications.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eFive and twelve features were selected for predicting obesity under Chinese and WHO standards, respectively. Age, maternal BMI, paternal BMI, screen time, and birth weight were consistently important across both standards. The neural network model performed best under Chinese standards (AUC = 0.7352), while XGBoost performed best under WHO standards (AUC = 0.7358). SHAP analysis provided global and local interpretations of feature contributions. User-friendly online prediction tools were developed and made publicly available.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThis study developed interpretable machine learning models that effectively predict childhood and adolescent obesity using a minimal set of clinically relevant features. The deployed tools facilitate individualized risk assessment and may support targeted prevention strategies.\u003c/p\u003e","manuscriptTitle":"Interpretable Machine Learning Models for Childhood and Adolescent Obesity Prediction According to Chinese and WHO Standards: A Two-Wave Cross-Sectional Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-04 05:19:14","doi":"10.21203/rs.3.rs-7617689/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"aead0c95-219e-468d-90cb-5939a87ee0dc","owner":[],"postedDate":"February 4th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":62192433,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":62192434,"name":"Health sciences/Diseases"},{"id":62192435,"name":"Health sciences/Health care"},{"id":62192436,"name":"Health sciences/Medical research"},{"id":62192437,"name":"Health sciences/Risk factors"}],"tags":[],"updatedAt":"2026-04-01T08:58:47+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-04 05:19:14","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7617689","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7617689","identity":"rs-7617689","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}