Machine Learning Integration of Serial Blood Biomarkers Enhances Cognitive Decline Prediction in Early Parkinson's Disease

Machine Learning Integration of Serial Blood Biomarkers Enhances Cognitive Decline Prediction in Early Parkinson's Disease | 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 Machine Learning Integration of Serial Blood Biomarkers Enhances Cognitive Decline Prediction in Early Parkinson's Disease Raziyeh Mohammadi, Samuel Y. E. Ng, Jayne Y. Tan, Adeline S. L. Ng, and 12 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7107548/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 25 Feb, 2026 Read the published version in npj Parkinson's Disease → Version 1 posted 10 You are reading this latest preprint version Abstract Cognitive decline is a major non-motor complication in early Parkinson’s disease (PD), but predicting its progression remains challenging. Using data from 193 participants in the Early Parkinson’s Disease Longitudinal Singapore (PALS) cohort, we evaluated whether longitudinal blood biomarkers—neurofilament light chain (NfL) and total tau (t-tau)—could improve prediction of cognitive decline, defined as a one-point annual or sustained two-year drop in Montreal Cognitive Assessment scores. We applied three variable selection methods and five machine learning models across seven feature sets. Overall, 23% of participants experienced cognitive decline over five years. The XGBoost model trained on Random Forest–selected variables achieved the highest performance (AUC = 0.806), a substantial improvement over the baseline-only model (AUC = 0.560). Key predictors included diastolic blood pressure and summaries of t-tau and NfL. Time-varying biomarkers improved predictions over baseline data alone, supporting their integration with machine learning for early cognitive risk assessment in PD. Health sciences/Biomarkers Health sciences/Neurology Biological sciences/Neuroscience Figures Figure 1 Figure 2 Introduction Parkinson disease (PD) is the second most common neurodegenerative disorder, affecting over 10 million people globally. With aging populations, its prevalence is expected to double by 2040, increasing burdens on healthcare systems and caregivers. Although PD is primarily defined by motor symptoms such as bradykinesia, rigidity, and tremor, non-motor features—particularly cognitive impairment—contribute significantly to morbidity [ 1 – 4 ]. Cognitive decline affects up to 80% of patients, with many progressing from mild cognitive impairment (PD-MCI) to Parkinson’s disease dementia (PDD) within a few years of diagnosis [ 5 – 10 ]. Cognitive deficits in PD have serious clinical and societal impacts, including reduced quality of life, loss of independence, and higher rates of institutionalization. Healthcare costs for patients with PDD are twice those without dementia, and caregivers face increased psychological and financial strain. Predicting cognitive decline remains challenging due to heterogeneous disease progression. Early identification of at-risk patients is crucial for timely intervention, counseling, and improving clinical trial design [ 7 , 11 – 14 ]. Current prognostic tools poorly predict cognitive outcomes in PD. Clinical scales like the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) focus on motor symptoms and offer limited insight into cognitive decline. While Cerebrospinal fluid (CSF) biomarkers and neuroimaging provide some prognostic value, they are invasive, costly, and impractical for routine use [ 15 , 16 ]. In contrast, blood-based biomarkers are minimally invasive, cost-effective, and feasible for repeated assessments [ 17 , 18 ]. Neurofilament light chain (NfL), a marker of axonal injury, correlates strongly with CSF levels and is linked to disease severity and cognitive decline in PD [ 19 – 23 ]. Similarly, total tau (t-tau) and phosphorylated tau181 (p-tau181) have been associated with cognitive dysfunction [ 24 – 26 ]. Prior studies have reported elevated plasma NfL and t-tau in individuals with PD who exhibit early cognitive deficits, and combinations of plasma α-synuclein, p-tau181, and Aβ-40 have been proposed as potential biomarkers [ 26 – 34 ]. However, the vast majority of these investigations rely on cross-sectional data, limiting their predictive power. Longitudinal biomarker trajectories, such as changes in NfL and t-tau, may better reflect disease progression [ 35 , 36 ], yet few studies have evaluated their predictive utility in early PD or combined them with clinical data to improve prognostic accuracy [ 37 ]. Advances in machine learning (ML) have transformed neurodegeneration research by enabling the analysis of complex, high-dimensional data [ 38 – 40 ]. ML can detect non-linear patterns and interactions often missed by traditional methods[ 41 ]. Prior work using data from the Parkinson’s Progression Markers Initiative (PPMI) has shown that ML models can achieve high accuracy in predicting disease progression by incorporating multimodal data, including motor scores, biospecimens, and patient-reported outcomes [ 42 ]. However, the utility of ML for predicting cognitive decline in early PD using time-varying blood biomarkers remains underexplored. Building on our prior work predicting cognitive decline in early PD using baseline features (AUC = 0.93) [ 43 ], we now examine whether adding longitudinal plasma NfL and t-tau measurements improves prediction. Using data from the Early Parkinson’s Disease Longitudinal Singapore (PALS) cohort, we apply multiple ML algorithms to model cognitive decline over five years. Our objectives are to: (1) assess the added value of longitudinal biomarkers, (2) identify key predictors using advanced feature selection, and (3) develop a clinically relevant risk prediction tool. Methods Study Design, Participants, and Variables This prospective observational study used data from the PALS cohort, which tracks clinical, cognitive, and biological changes in early PD. Participants were enrolled from specialist clinics in Singapore (2014–2018). Inclusion criteria were PD diagnosis within 1 year, symptom onset within 2 years, and meeting the National Institute of Neurological Disorders and Stroke criteria. Participants required at least six years of education. Exclusion criteria included comorbidities likely to confound outcomes or hinder follow-up (e.g., malignancy, unstable cardiovascular disease) [ 44 , 45 ]. Of 214 participants recruited, 193 completed baseline assessments and had at least three years of cognitive follow-up, comprising the final analytic cohort. Informed consent was obtained; the study was approved by the Singapore Health Services Centralized Institutional Review Board. Participants underwent detailed annual evaluations over five years, including clinical, cognitive, and laboratory assessments. Cognitive performance was measured using the Montreal Cognitive Assessment (MoCA). Venous blood samples were collected at baseline, year 3, and year 5. Primary biomarkers—NfL and t-tau—were quantified using ultra-sensitive single-molecule array (Simoa) assays. For these biomarkers, four summary features (minimum, maximum, mean, and standard deviation) were computed across time points to capture both levels and variability. Additional baseline biomarkers included suppression of tumorigenicity 2 (ST2), apolipoprotein E (APOE) genotype, p-tau181, and alpha-synuclein gene promoter repeat length (REP1). The primary outcome was cognitive decline, defined as either a ≥ 1-point annual decrease in MoCA score or a ≥ 1-point drop sustained across two consecutive years during the five-year follow-up. This threshold was selected to reflect subtle but clinically meaningful cognitive changes. Baseline demographic characteristics included age, sex, body mass index (BMI), years of education, smoking history, alcohol use, and consumption of coffee and tea. Clinical features comprised Hoehn and Yahr stage (HY), baseline MoCA score, MDS-UPDRS Part III score, systolic and diastolic blood pressure (SBP and DBP) measured in both lying and standing positions, and comorbidities including hypertension, diabetes mellitus, hypertension and hyperlipidemia. Statistical Analysis Missing data were addressed using a random forest (RF)–based imputation method [ 46 ]. All continuous variables were examined for clinical interpretability, and when appropriate, were dichotomized using clinically relevant thresholds. In the absence of established cutoffs, the Youden index was used to derive optimal cut-points from ROC curves to maximize sensitivity and specificity [ 47 ]. Descriptive statistics were used to summarize all variables. Continuous variables were presented as mean (standard deviation) or median (interquartile range), and categorical variables as counts and percentages. Associations between individual predictors and cognitive decline were assessed using univariate logistic regression, reporting odds ratios (ORs) with 95% confidence intervals (CIs). To identify the most informative predictors, we applied three complementary variable selection methods. First, RF importance scores were used to rank variables by mean decrease in Gini impurity, retaining only those above the average score. Second, the Shapley Variable Importance Cloud (SHAPvIC) method quantified each variable’s contribution using SHAP values; variables with 95% prediction intervals entirely above zero were selected. Third, we applied gain-based feature importance from Extreme Gradient Boosting (XGBoost), keeping variables with scores above the mean. This triangulated approach enhanced both statistical robustness and consistency across methods. For predictive modeling, we trained five supervised ML algorithms: RF, neural network (NN), k-nearest neighbors (KNN), XGBoost, and Light Gradient Boosting Machine (LightGBM). Each algorithm was evaluated across seven different feature sets: time-varying biomarkers alone; baseline variables alone; a combination of both; and subsets derived from each of the three variable selection strategies (including a reduced version of SHAPvIC for parsimony). The dataset was randomly split (80:20) into training and test sets, stratified by cognitive decline status. Hyperparameters were optimized via five-fold cross-validation on the training set. For RF, we tuned the number of trees and number of predictors considered at each split. For XGBoost and LightGBM, we optimized learning rate (0.01–0.1), number of boosting rounds (50–500), maximum tree depth (3–7), and L1/L2 regularization terms (range: 0–1). NN models varied in hidden layer size (1–5) and applied weight decay (0–0.1), while the optimal number of neighbors (k = 1–10) was selected for KNN models. Model performance was assessed on the held-out test set using multiple metrics. Discrimination was evaluated via AUC with 95% CIs from 1,000 bootstrap resamples. Additional metrics included sensitivity, specificity, accuracy, precision, recall, F1 score, and Brier score. The F1 score, the harmonic mean of precision and recall, balances false positives and negatives. The Brier score quantifies the accuracy of probabilistic predictions. Sensitivity and specificity were calculated at the optimal threshold via Youden’s index. Calibration was assessed using decile-based plots comparing observed and predicted event rates across 10 bins. All analyses were performed using R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was set at a two-sided P Value < 0.05. Results A total of 193 participants with early PD were included. At baseline, the mean age was 63.8 years (SD = 8.9), 58% were male, and the median education duration was 11 years (IQR = 8–15). The mean MoCA score was 25.4 (SD = 2.8). Disease severity was mild to moderate: 62% in HY stage 1, 30% in stage 2, and 8% in stage 3. Over five years, 44 participants (22.8%) experienced cognitive decline (Table 1 ). Higher education (≥ 10 years) was protective (OR = 0.4; 95% CI: 0.2–0.7; P = .0006), while elevated blood pressure increased risk: lying DBP (OR = 2.6; 95% CI: 1.3–5.2; P = .009), standing DBP (OR = 2.1; 95% CI: 1.1–4.3; P = .045), and lying SBP (OR = 2.1; 95% CI: 1.1–4.3; P = .045) (Table 1 ). Table 1 Summary statistics of demographic, clinical assessments, blood biomarkers, and their associations with progression outcomes using univariate logistic regression. Total No Progression Progression OR (95% CIs) P Value N = 193 N = 149 N = 44 Demographic Characteristics Male Gender 112 (58.0%) 85 (57.0%) 27 (61.4%) 1.2 (0.6, 2.4) 0.737 Smoker 56 (29.0%) 43 (28.9%) 13 (29.5%) 1.0 (0.5, 2.1) 1.000 Years of education (≥ 10 years) 135 (69.9%) 112 (75.2%) 23 (52.3%) 0.4 (0.2, 0.7) 0.006 Tea drinking 180 (93.3%) 138 (92.6%) 42 (95.5%) 1.6 (0.4, 11.5) 0.736 Coffee drinking 175 (90.7%) 135 (90.6%) 40 (90.9%) 1.0 (0.3, 3.8) 1.000 Alcohol drinking 125 (64.8%) 101 (67.8%) 24 (54.5%) 0.6 (0.3, 1.1) 0.151 BMI (> 25 kg/m2) 64 (33.2%) 48 (32.2%) 16 (36.4%) 1.2 (0.6, 2.4) 0.740 Age (> 65 years) 97 (50.3%) 72 (48.3%) 25 (56.8%) 1.4 (0.7, 2.8) 0.413 Clinical Assessments Lying SBP (≥ 140 mmHg) 95 (49.2%) 67 (45.0%) 28 (63.6%) 2.1 (1.1, 4.3) 0.045 Lying DBP (≥ 80 mmHg) 67 (34.7%) 44 (29.5%) 23 (52.3%) 2.6 (1.3, 5.2) 0.009 Standing SBP (≥ 140 mmHg) 85 (44.0%) 63 (42.3%) 22 (50.0%) 1.4 (0.7, 2.7) 0.463 Standing DBP (≥ 80 mmHg) 95 (49.2%) 67 (45.0%) 28 (63.6%) 2.1 (1.1, 4.3) 0.045 Diabetes mellitus 31 (16.1%) 25 (16.8%) 6 (13.6%) 0.8 (0.3, 2.0) 0.791 Hypertension 88 (45.6%) 68 (45.6%) 20 (45.5%) 1.0 (0.5, 2.0) 1.000 Hyperlipidemia 92 (47.7%) 73 (49.0%) 19 (43.2%) 0.8 (0.4, 1.6) 0.613 MoCA 26 [23.0, 28.0] 26 [23.0, 28.0] 26 [23.0, 28.0] 1.0 (0.9, 1.1) 0.771 Total motor score 20.0 [15.0; 26.0] 19.0 [15.0; 26.0] 22.0 [17.0; 29.0] 1.0 (1.0, 1.1) 0.062 HY 2.00 [1.0; 3.0] 2.00 [1.50; 2.0] 2.00 [2.00; 2.0] 2.0 (0.8, 4.8) 0.112 Blood Biomarkers APOE4 (Non-carriers) 153 (79.3%) 120 (80.5%) 33 (75.0%) 0.7 (0.3, 1.7) 0.559 Rep 1 (Short) 88 (45.6%) 66 (44.3%) 22 (50.0%) 1.3 (0.6, 2.5) 0.620 ST2 11,596 [8747; 14,830] 11,534 [8,402; 14,854] 12611 [9,433; 14,813] 1.0 (1.0, 1.0) 0.375 NfL 13.7 [10.1; 18.9] 13.9 [10.2; 18.7] 13.3 [9.9; 21.7] 1.0 (1.0, 1.1) 0.702 t-tau 1.17 [0.9; 1.5] 1.1 [0.9; 1.6] 1.3 [0.9; 1.5] 1.3 (0.9, 1.8) 0.350 p-tau181 20.3 [15.7; 24.8] 20.5 [15.4; 24.3] 20.1 [15.8; 28.9] 1.0 (1.0, 1.1) 0.666 Data are expressed as frequency (%) or median (quartile); P values are from univariate logistic regression models assessing the association of each variable with cognitive decline progression. Abbreviations: N: number, OR: odds ratio, CIs: confidence intervals, BMI: body mass index, SBP: systolic blood pressure, DBP: diastolic blood pressure, MoCA: Montreal Cognitive Assessment, HY: Hoehn and Yahr scale, APOE: apolipoprotein E, REP1: alpha-synuclein gene promoter, ST2: suppression of tumorigenicity 2, NfL: neurofilament light chain, t-tau: total tau, p-tau181: phosphorylated tau at threonine 181. To identify the most predictive variables, thirty candidate features were assessed using three feature selection methods. The RF approach identified twelve variables, including years of education, lying and standing DBP, and four summary statistics (minimum, maximum, mean, and standard deviation) for both NfL and t-tau, along with baseline p-tau181 levels (Fig. 1, top row, left). The ShapleyVIC method selected fourteen predictors, which included t-tau (minimum, maximum, mean, and standard deviation), NfL (minimum and mean), as well as baseline MoCA score, hyperlipidemia, coffee and alcohol consumption, lying and standing DBP, p-tau181, and gender (Fig. 1, top row, right). The XGBoost-based method identified eleven key features, including the full set of summary metrics for both NfL and t-tau, lying and standing DBP, and years of education (Fig. 1, bottom row). Notably, eight variables—minimum, maximum, mean, and standard deviation of t-tau; minimum and mean NfL; and lying and standing DBP—were consistently selected across all three methods. ML performance varied across algorithms and feature sets (Table 2). Models using longitudinal biomarker summaries (Model 1) consistently outperformed those using only baseline variables (Model 2), with AUCs ranging from 0.640 to 0.763 vs 0.465 to 0.560, respectively. The XGBoost model using RF-selected variables (Model 4) achieved the highest AUC (0.806; 95% CI, 0.663–0.949) and the best overall performance, with accuracy of 0.730, sensitivity of 0.875, specificity of 0.690, F1 score of 0.583, and a Brier score of 0.140. The LightGBM model using time-varying biomarker summaries (Model 1) performed well (AUC = 0.763; 95% CI, 0.598–0.928), as did both LightGBM and XGBoost models using SHAPvIC-selected variables (Model 5; AUC = 0.698 for both). These models showed comparable sensitivity and specificity to the top-performing XGBoost model, though with slightly lower precision and calibration. Notably, the reduced SHAPvIC subset (Model 6), comprising six variables, achieved the highest specificity (0.862; 95% CI, 0.683–0.961) with XGBoost, highlighting the utility of parsimonious models in clinical prediction. Table 2. Performance of ML models across different feature sets. Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 RF AUC (95% CI) 0.763 (0.613, 0.912) 0.478 (0.201,0.755) 0.6056 (0.395, 0.817) 0.6853 (0.499, 0.871) 0.733 (0.520, 0.946) 0.5345 (0.299, 0.770) 0.580 (0.351, 0.808) Sensitivity (95% CI) 0.875 (0.473, 0.997) 0.500 (0.157, 0.843) 0.750 (0.349, 0.968) 0.750 (0.349, 0.968) 0.750 (0.349, 0.968) 0.500 (0.157, 0.843) 0.750 (0.349, 0.968) Specificity (95% CI) 0.655 (0.457, 0.821) 0.759 (0.565, 0.897) 0.483 (0.294, 0.675) 0.621 (0.423, 0.793) 0.690 (0.492, 0.847) 0.621 (0.423, 0.793) 0.483 (0.294, 0.675) Accuracy (95% CI) 0.703 (0.530, 0.841) 0.703 (0.530, 0.841) 0.540 (0.369, 0.705) 0.649 (0.475, 0.798) 0.703 (0.530, 0.841) 0.595 (0.421, 0.752) 0.540 (0.369, 0.705) Precision 0.4118 0.364 0.286 0.353 0.400 0.267 0.286 F1 Score 0.560 0.421 0.414 0.480 0.522 0.348 0.414 Log Loss 0.477 NE 0.529 0.514 0.478 0.567 0.612 Brier Score 0.157 0.2010 0.174 0.174 0.154 0.187 0.180 NN AUC (95% CI) 0.711 (0.542, 0.880) 0.534 (0.267, 0.802) 0.685 (0.497, 0.874) 0.668 (0.485, 0.852) 0.5948 (0.386, 0.804) 0.612 (0.370, 0.854) 0.655 (0.468, 0.843) Sensitivity (95% CI) 0.875 (0.473, 0.997) 0.625 (0.245, 0.915) 0.625 (0.245, 0.915) 0.875 (0.473, 0.997) 0.6250 (0.245, 0.915) 0.7500 (0.349, 0.968) 0.8750 (0.4735, 0.9968) Specificity (95% CI) 0.586 (0.389, 0.765) 0.517 (0.325, 0.705) 0.759 (0.565, 0.897) 0.517 (0.325, 0.705) 0.6897 (0.492, 0.847) 0.6552 (0.457, 0.821) 0.483 (0.294, 0.675) Accuracy (95% CI) 0.649 (0.475, 0.798) 0.5405 (0.369, 0.705) 0.730 (0.559, 0.862) 0.595 (0.421, 0.753) 0.676 (0.502, 0.820) 0.676 (0.502, 0.820) 0.568 (0.395, 0.729) Precision 0.368 0.263 0.417 0.333 0.357 0.375 0.318 F1 Score 0.518 0.370 0.500 0.483 0.454 0.500 0.467 Log Loss 0.477 1.046 0.775 0.634 0.809 0.529 0.556 Brier Score 0.159 0.232 0.239 0.214 0.254 0.172 0.194 KNN AUC (95% CI) 0.640 (0.440, 0.840) 0.560 (0.331, 0.789) 0.619 (0.408, 0.829) 0.640 (0.436, 0.844) 0.534 (0.338-0.731) 0.603 (0.433, 0.774) 0.608 (0.421, 0.794) Sensitivity (95% CI) 0.625 (0.245, 0.915) 0.625 (0.245, 0.915) 0.750 (0.349, 0.968) 0.625 (0.245, 0.915) 0.500 (0.157, 0.843) 0.500 (0.157, 0.843) 0.375 (0.085, 0.755) Specificity (95% CI) 0.655 (0.457, 0.821) 0.448 (0.264, 0.643) 0.517 (0.325, 0.706) 0.655 (0.457, 0.821) 0.517 (0.325, 0.706) 0.621 (0.423, 0.793) 0.724 (0.528, 0.873) Accuracy (95% CI) 0.6486 (0.475, 0.798) 0.487 (0.319, 0.656) 0.568 (0.395, 0.729) 0.649 (0.475, 0.798) 0.513 (0.344, 0.681) 0.595 (0.42, 0.752) 0.649 (0.475, 0.798) Precision 0.333 0.238 0.300 0.333 0.222 0.267 0.273 F1 Score 0.435 0.345 0.429 0.435 0.308 0.348 0.316 Log Loss Inf 0.547 0.614 0.506 Inf 0.512 0.502 Brier Score 0.351 0.177 0.206 0.167 0.298 0.173 0.168 XGBoost AUC (95% CI) 0.737 (0.546, 0.928) 0.517 (0.238, 0.797) 0.595 (0.412, 0.778) 0.806 (0.663, 0.949) 0.698 (0.459, 0.938) 0.647 (0.411, 0.882) 0.659 (0.478, 0.841) Sensitivity (95% CI) 0.750 (0.349, 0.968) 0.625 (0.245, 0.915) 0.875 (0.473, 0.997) 0.875 (0.473, 0.997) 0.625 (0.245, 0.915) 0.500 (0.157, 0.843) 0.875 (0.473, 0.997) Specificity (95% CI) 0.621 (0.423, 0.793) 0.690 (0.492, 0.847) 0.517 (0.325, 0.706) 0.690 (0.492, 0.847) 0.828 (0.642, 0.942) 0.862 (0.683, 0.961) 0.552 (0.357, 0.736) Accuracy (95% CI) 0.649 (0.475, 0.798) 0.676 (0.502, 0.820) 0.595 (0.421, 0.752) 0.730 (0.550, 0.862) 0.784 (0.618, 0.902) 0.784 (0.618, 0.902) 0.622 (0.448, 0.775) Precision 0.353 0.357 0.333 0.437 0.500 0.500 0.350 F1 Score 0.480 0.454 0.483 0.583 0.556 0.500 0.500 Log Loss 0.451 0.9159 0.565 0.417 0.474 0.496 0.548 Brier Score 0.146 0.230 0.189 0.140 0.152 0.160 0.181 LightGBM AUC (95% CI) 0.763 (0.598, 0.928) 0.465 (0.191, 0.740) 0.681 (0.490, 0.872) 0.728 (0.535, 0.922) 0.698 (0.476, 0.920) 0.465 (0.228, 0.703) 0.690 (0.501, 0.878) Sensitivity (95% CI) 0.750 (0.349, 0.968) 0.500 (0.157, 0.843) 0.625 (0.245, 0.915) 0.750 (0.349, 0.968) 0.625 (0.245, 0.915) 0.500 (0.157, 0.843) 0.875 (0.473, 0.997) Specificity (95% CI) 0.759 (0.565, 0.897) 0.690 (0.492, 0.847) 0.724 (0.528, 0.873) 0.690 (0.492, 0.847) 0.828 (0.642, 0.942) 0.690 (0.492, 0.847) 0.552 (0.357, 0.736) Accuracy (95% CI) 0.757 (0.588, 0.882) 0.649 (0.475, 0.797) 0.703 (0.530, 0.841) 0.703 (0.530, 0.841) 0.784 (0.618, 0.902) 0.6486 (0.475, 0.798) 0.622 (0.448, 0.775) Precision 0.461 0.308 0.385 0.400 0.500 0.308 0.350 F1 Score 0.571 0.381 0.476 0.522 0.556 0.381 0.500 Log Loss 0.567 0.908 0.5555 0.537 0.488 0.526 0.624 Brier Score 0.182 0.234 0.187 0.173 0.158 0.171 0.203 Abbreviations: NE: Not estimable due to infinite log loss arising from zero predicted probabilities. Evaluation was conducted using the test dataset from an 80/20 train-test split, with hyperparameters optimized via 10-fold cross-validation on the training set. Model 1: Descriptive statistics (mean, SD, Min, Max) of eight time-varying biomarkers. Model 2: 24 baseline variables. Model 3: Combined time-varying biomarkers and baseline variables (30 features). Model 4: Top 12 variables selected by RF (years of education; standing and lying DBP; four summary statistics (minimum, maximum, mean, Standard deviation) of both NfL and t-tau; and pTau181). Model 5: Top 14 variables identified using ShapleyVIC (four summary statistics of t-tau; minimum and mean NfL; standing and lying DBP; coffee and alcohol consumption; gender; hyperlipidemia; pTau181; and MoCA score). Model 6: Reduced set of 6 variables from ShapleyVIC-selected variables (maximum, mean, and Standard deviation of t-tau; minimum and mean NfL; and standing DBP). Model 7: Top 11 variables from XGBoost (four summary statistics of both t-tau and NfL; years of education; standing DBP; and lying DBP). Fig. 2 shows calibration results for Model 4, based on top predictors selected via RF. XGBoost and LightGBM demonstrated the best calibration, with mean absolute errors <0.05 and close alignment to the 45-degree reference line, indicating well-calibrated predictions. The RF model was reasonably calibrated but showed deviation in lower predicted probabilities. In contrast, KNN and NN models were poorly calibrated, underestimating risk in high-risk individuals and overestimating in low-risk cases. In the best-performing model—XGBoost using RF-selected variables—key predictors included years of education, lying and standing DBP, four summary statistics (minimum, maximum, mean, and standard deviation) for both NfL and t-tau, and p-tau181 levels. SHAP visualizations showed that elevated and variable t-tau levels strongly influenced high-risk classification. Supine and upright DBP ≥90 mmHg also contributed substantially. While NfL features were associated with cognitive decline, they were less frequently among the top three predictors. Overall, incorporating longitudinal biomarker summaries—especially dynamic t-tau features—improved predictive performance and enabled more parsimonious models suitable for clinical application. Discussion In this prospective study of early PD, integrating longitudinal blood biomarkers—particularly t-tau and NfL—significantly enhanced cognitive decline prediction using ML. The XGBoost model with RF-selected features achieved the best performance (AUC = 0.806), a substantial improvement over the baseline-only model (AUC = 0.560). Across algorithms, dynamic changes in t-tau and DBP emerged as the most consistent predictors. The consistent identification of t-tau features highlights tau pathology's role in PD-related cognitive decline. Notably, t-tau variability was a strong predictor, indicating temporal fluctuations may reflect meaningful biological processes like axonal damage or inflammation. Although CSF biomarkers are informative, blood-based assays offer minimal invasiveness and increasing reliability with technologies like Simoa. Our results support using serial plasma t-tau as a prognostic tool in PD, enabling longitudinal monitoring of neurodegeneration. Patients with rising or fluctuating t-tau levels could be candidates for further assessment or early intervention. The inclusion of NfL as a predictive biomarker also corroborates prior evidence of its value [48, 49]. NfL reflects axonal injury and has been associated with faster motor and cognitive progression in PD [35, 50-55]. However, its predictive strength in our study was somewhat less robust than that of t-tau, suggesting that while NfL may capture broader neurodegenerative processes, tau-specific mechanisms may be more intimately tied to cognitive trajectories. These distinctions could be important in future therapeutic trials, where stratifying patients by biomarker phenotype may help tailor interventions or identify treatment responders. Beyond molecular biomarkers, our study reinforces the prognostic importance of DBP—both in lying and standing positions—as a consistent clinical predictor of cognitive decline. The association of elevated DBP with two-fold increased risk has significant clinical implications [56-58]. Autonomic dysfunction is common in PD and can lead to labile blood pressure patterns, including supine hypertension and orthostatic hypotension. These fluctuations may impair cerebral autoregulation, reduce perfusion to vulnerable brain regions, and accelerate neurodegeneration. Current PD guidelines focus on motor symptoms, but our findings support routine blood pressure monitoring to identify vascular risk and guide early intervention. Importantly, blood pressure represents a modifiable risk factor. Interventions such as midodrine, fludrocortisone, or domperidone may be used to address hypotension, while antihypertensives including ACE inhibitors or calcium channel blockers could be considered for elevated supine DBP, ideally under close monitoring to avoid worsening orthostatic symptoms. In patients with high DBP and fluctuating tau profiles, a multidisciplinary approach involving neurology, cardiology, and internal medicine may be warranted. Incorporating vascular health into PD management may not only reduce cognitive risk but also improve cardiovascular outcomes in this aging population. Another key finding was the protective effect of education, supporting the cognitive reserve hypothesis. Participants with ≥10 years of formal education had a 58% lower odds of cognitive decline than those with less education. While not modifiable retrospectively, educational attainment may aid in risk stratification. Individuals with limited education could benefit from closer cognitive monitoring, targeted interventions, and referral to neuropsychology. They may also be prioritized for emerging strategies to preserve cognitive function, including lifestyle modification, cognitive training, and digital therapeutics. Our results also point to the potential value of deploying ML-based prediction tools in clinical workflows. By integrating time-varying biomarker data, clinical measures, and demographic factors, ML algorithms can generate individualized risk scores that support early identification and proactive care. For instance, a PD patient with high DBP and rising t-tau levels could be flagged for cognitive rehabilitation or clinical trial referral, while a low-risk individual might focus on motor symptom management. These tools could be embedded in electronic health records (EHRs), providing decision support to clinicians and improving consistency in care delivery. Furthermore, ML-derived insights could enhance patient-provider communication, allowing neurologists to frame prognosis in more concrete terms, support advance care planning, and tailor surveillance strategies. In clinical trial design, predictive models like ours can support enrichment strategies to boost statistical power and reduce sample sizes. Given the heterogeneity of PD progression, selecting high-risk individuals based on longitudinal tau and DBP dynamics may enhance trial efficiency and increase the likelihood of detecting treatment effects. This approach is particularly valuable for emerging disease-modifying therapies, such as anti-tau and anti-synuclein agents, where early intervention is key. Our findings also align with broader evidence suggesting that mixed pathology is common in PD, particularly in those with cognitive impairment [59, 60]. While PD is defined pathologically by Lewy bodies composed of α-synuclein, many patients with dementia show concurrent tau and amyloid pathology on autopsy or PET imaging [61, 62]. The observed importance of tau dynamics in our study supports this concept and may prompt reconsideration of PD as a disease spectrum rather than a single pathology [63]. This perspective opens the door to cross-disease therapeutic strategies and biomarker-guided personalization of care, particularly for those who meet overlapping criteria for PD, dementia with Lewy bodies, or AD [64, 65]. From a mechanistic standpoint, the interplay between vascular dysregulation and tau-driven neurodegeneration is particularly noteworthy. Chronic cerebrovascular dysfunction can disrupt glymphatic clearance of toxic proteins, promote blood-brain barrier breakdown, and create a permissive environment for tau aggregation. Conversely, tau pathology may exacerbate vascular reactivity deficits through endothelial dysfunction or oxidative stress. This bidirectional model suggests that targeting both pathways may offer synergistic benefits. For example, interventions aimed at reducing vascular stiffness or enhancing neurovascular coupling might complement disease-modifying therapies targeting tau aggregation or phosphorylation. Despite its strengths, this study has limitations. While we used rigorous internal validation, external validation in independent cohorts is needed to confirm generalizability. The primarily Asian composition of the PALS cohort may limit applicability to other populations. Biomarkers were measured at only three time points; more frequent sampling could enhance trajectory analysis. Finally, although we focused on interpretable features and avoided overfitting, future studies should assess whether adding biomarkers of synaptic dysfunction, neuroinflammation, or α-synuclein activity improves prediction. Future research should prioritize replication in multiethnic cohorts, integration of imaging biomarkers (e.g., tau PET, diffusion MRI), and development of practical risk calculators for clinical use. Cost-effectiveness studies will be essential to evaluate the feasibility of widespread biomarker monitoring. As high-sensitivity blood assays become more accessible, their implementation will be facilitated by advances in digital health, cloud analytics, and decentralized data sharing. Conclusions This study shows that integrating longitudinal biomarker dynamics, particularly t-tau, with ML significantly enhances prediction of cognitive decline in early PD. These findings support the use of regular biomarker monitoring, vascular risk management, and predictive tools to enable personalized care and improve trial efficiency—marking a key step toward precision neurology in PD. Declarations Author Contributions: Conceptualization, R.M., A.S.L.N., E.-K.T., L.C.S.T., and S.E.S.; Methodology, R.M., W.G., and S.E.S.; Software, R.M. and S.E.S.; Validation, R.M. and S.E.S.; Formal Analysis, R.M. and S.E.S.; Investigation, R.M. and S.E.S.; Resources, S.N., J.Y.T., A.S.L.N., X.D., X.C., D.L.H., S.N., Z.X., K.-Y.T., W.-L.A., E.-K.T., L.C.S.T., and S.E.S.; Data Curation, S.N., J.Y.T., A.S.L.N., X.D., X.C., and S.E.S.; Writing—Original Draft Preparation, R.M. and S.E.S.; Writing—Review and Editing, R.M., S.E.S., J.Y.T., A.S.L.N., X.D., X.C., D.L.H., S.N., Z.X., K.-Y.T., W.-L.A., E.-K.T., L.C.S.T., W.G., M.L., and S.E.S.; Visualization, R.M. and S.E.S.; Supervision, A.S.L.N., E.-K.T., L.C.S.T., W.G., M.L., and S.E.S.; Project Administration, R.M., S.Y.E.N., and J.Y.T.; Funding Acquisition, E.-K.T., L.C.S.T., and S.E.S. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the Singapore Ministry of Health’s National Medical Research Council (MOH-OFLCG18May-0002, MOH-CSAINV21-0005, CNIG22jul-0004). Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of SingHealth (IRB reference number: CIRB 2019-2433) and National University of Singapore (IRB reference number: NUS-IRB-2022-899). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The study data will be made available upon reasonable request to the corresponding author. The data are not publicly available due to privacy and ethical concerns. Conflicts of Interest: The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results. References Hely, M. A.; Reid, W. G. J.; Adena, M. A.; Halliday, G. M.; Morris, J. G. L., The Sydney multicenter study of Parkinson's disease: the inevitability of dementia at 20 years. Movement disorders 2008, 23 (6), 837-844. Simuni, T.; Sethi, K., Nonmotor manifestations of Parkinson's disease. Annals of Neurology: Official Journal of the American Neurological Association and the Child Neurology Society 2008, 64 (S2), S65-S80. Riedel, O.; Klotsche, J.; Spottke, A.; Deuschl, G.; Förstl, H.; Henn, F.; Heuser, I.; Oertel, W.; Reichmann, H.; Riederer, P., Frequency of dementia, depression, and other neuropsychiatric symptoms in 1,449 outpatients with Parkinson’s disease. Journal of neurology 2010, 257 , 1073-1082. Deng, X.; Saffari, S. E.; Xiao, B.; Ng, S. Y. E.; Chia, N.; Choi, X.; Heng, D. L.; Xu, Z.; Tay, K.-Y.; Au, W.-L., Disease progression in Parkinson’s disease patients with mild cognitive impairment: 5-year longitudinal study from the early Parkinson’s disease longitudinal Singapore (PALS) cohort. Aging (Albany NY) 2024, 16 (15), 11491. Xu, F.; Huang, H.; Feng, J.; Shen, Q.; Bao, Y.; Zhang, D.; Xu, Y., Cerebrospinal fluid tau and disease progression in early Parkinson's disease: an 8-year longitudinal study. J Neurol 2024, 272 (1), 61. Williams-Gray, C. H.; Mason, S. L.; Evans, J. R.; Foltynie, T.; Brayne, C.; Robbins, T. W.; Barker, R. A., The CamPaIGN study of Parkinson's disease: 10-year outlook in an incident population-based cohort. Journal of Neurology, Neurosurgery & Psychiatry 2013, 84 (11), 1258-1264. Greenland, J. C.; Williams‐Gray, C. H.; Barker, R. A., The clinical heterogeneity of Parkinson's disease and its therapeutic implications. European Journal of Neuroscience 2019, 49 (3), 328-338. Roheger, M.; Kalbe, E.; Liepelt-Scarfone, I., Progression of cognitive decline in Parkinson’s disease. Journal of Parkinson's disease 2018, 8 (2), 183-193. Deng, X.; Saffari, S. E.; Xiao, B.; Ng, S. Y. E.; Chia, N.; Choi, X.; Heng, D. L.; Ng, E.; Xu, Z.; Tay, K.-Y., Disease Progression of Data-Driven Subtypes of Parkinson’s Disease: 5-Year Longitudinal Study from the Early Parkinson’s Disease Longitudinal Singapore (PALS) Cohort. Journal of Parkinson’s Disease 2024, 14 (5), 1051-1059. Fang, C.; Lv, L.; Mao, S.; Dong, H.; Liu, B., Cognition deficits in Parkinson’s disease: mechanisms and treatment. Parkinson’s disease 2020, 2020 (1), 2076942. Pressley, J. C.; Louis, E. D.; Tang, M. X.; Cote, L.; Cohen, P. D.; Glied, S.; Mayeux, R., The impact of comorbid disease and injuries on resource use and expenditures in parkinsonism. Neurology 2003, 60 (1), 87-93. Shen, J.; Amari, N.; Zack, R.; Skrinak, R. T.; Unger, T. L.; Posavi, M.; Tropea, T. F.; Xie, S. X.; Van Deerlin, V. M.; Dewey Jr, R. B., Plasma MIA, CRP, and albumin predict cognitive decline in Parkinson's disease. Annals of neurology 2022, 92 (2), 255-269. Mattsson, N.; Andreasson, U.; Zetterberg, H.; Blennow, K.; Alzheimer’s Disease Neuroimaging, I., Association of plasma neurofilament light with neurodegeneration in patients with Alzheimer disease. JAMA neurology 2017, 74 (5), 557-566. Aarsland, D.; Creese, B.; Politis, M.; Chaudhuri, K. R.; Ffytche, D. H.; Weintraub, D.; Ballard, C., Cognitive decline in Parkinson disease. Nature Reviews Neurology 2017, 13 (4), 217-231. Oizumi, H.; Hasegawa, T.; Kawahata, I.; Sekimori, T.; Totsune, T.; Sugimura, Y.; Baba, T.; Fukunaga, K.; Takeda, A., Associations among blood biomarkers, clinical subtypes, and prognosis in Parkinson’s disease. Clinical Parkinsonism & Related Disorders 2025, 12 , 100313. Martin, F. L.; Williamson, S. J. M.; Paleologou, K. E.; Allsop, D.; El-Agnaf, O. M. A., α-Synuclein and the pathogenesis of Parkinson's disease. Protein and Peptide Letters 2004, 11 (3), 229-237. Parnetti, L.; Gaetani, L.; Eusebi, P.; Paciotti, S.; Hansson, O.; El-Agnaf, O.; Mollenhauer, B.; Blennow, K.; Calabresi, P., CSF and blood biomarkers for Parkinson's disease. The Lancet Neurology 2019, 18 (6), 573-586. Youssef, P.; Hughes, L.; Kim, W. S.; Halliday, G. M.; Lewis, S. J. G.; Cooper, A.; Dzamko, N., Evaluation of plasma levels of NFL, GFAP, UCHL1 and tau as Parkinson's disease biomarkers using multiplexed single molecule counting. Sci Rep 2023, 13 (1), 5217. Khalil, M.; Teunissen, C. E.; Otto, M.; Piehl, F.; Sormani, M. P.; Gattringer, T.; Barro, C.; Kappos, L.; Comabella, M.; Fazekas, F., Neurofilaments as biomarkers in neurological disorders. Nature Reviews Neurology 2018, 14 (10), 577-589. Bäckström, D. C.; Domellöf, M. E.; Linder, J.; Olsson, B.; Öhrfelt, A.; Trupp, M.; Zetterberg, H.; Blennow, K.; Forsgren, L., Cerebrospinal fluid patterns and the risk of future dementia in early, incident Parkinson disease. JAMA neurology 2015, 72 (10), 1175-1182. Lerche, S.; Wurster, I.; Röben, B.; Zimmermann, M.; Machetanz, G.; Wiethoff, S.; Dehnert, M.; Rietschel, L.; Riebenbauer, B.; Deuschle, C., CSF NFL in a longitudinally assessed PD cohort: age effects and cognitive trajectories. Movement Disorders 2020, 35 (7), 1138-1144. Bacioglu, M.; Maia, L. F.; Preische, O.; Schelle, J.; Apel, A.; Kaeser, S. A.; Schweighauser, M.; Eninger, T.; Lambert, M.; Pilotto, A., Neurofilament light chain in blood and CSF as marker of disease progression in mouse models and in neurodegenerative diseases. Neuron 2016, 91 (1), 56-66. Lin, C.-H.; Li, C.-H.; Yang, K.-C.; Lin, F.-J.; Wu, C.-C.; Chieh, J.-J.; Chiu, M.-J., Blood NfL: a biomarker for disease severity and progression in Parkinson disease. Neurology 2019, 93 (11), e1104-e1111. Mollenhauer, B.; Bibl, M.; Wiltfang, J.; Steinacker, P.; Ciesielczyk, B.; Neubert, K.; Trenkwalder, C.; Otto, M., Total tau protein, phosphorylated tau (181p) protein, β-amyloid1–42, and β-amyloid1–40 in cerebrospinal fluid of patients with dementia with Lewy bodies. De Gruyter: 2006. Siderowf, A.; Xie, S. X.; Hurtig, H.; Weintraub, D.; Duda, J.; Chen-Plotkin, A.; Shaw, L. M.; Van Deerlin, V.; Trojanowski, J. Q.; Clark, C., CSF amyloid β 1-42 predicts cognitive decline in Parkinson disease. Neurology 2010, 75 (12), 1055-1061. Chen, N.-C.; Chen, H.-L.; Li, S.-H.; Chang, Y.-H.; Chen, M.-H.; Tsai, N.-W.; Yu, C.-C.; Yang, S.-Y.; Lu, C.-H.; Lin, W.-C., Plasma levels of α-synuclein, Aβ-40 and T-tau as biomarkers to predict cognitive impairment in Parkinson’s disease. Frontiers in aging neuroscience 2020, 12 , 112. Aamodt, W. W.; Waligorska, T.; Shen, J.; Tropea, T. F.; Siderowf, A.; Weintraub, D.; Grossman, M.; Irwin, D.; Wolk, D. A.; Xie, S. X.; Trojanowski, J. Q.; Shaw, L. M.; Chen-Plotkin, A. S., Neurofilament Light Chain as a Biomarker for Cognitive Decline in Parkinson Disease. Mov Disord 2021, 36 (12), 2945-2950. Ma, L. Z.; Zhang, C.; Wang, H.; Ma, Y. H.; Shen, X. N.; Wang, J.; Tan, L.; Dong, Q.; Yu, J. T., Serum Neurofilament Dynamics Predicts Cognitive Progression in de novo Parkinson's Disease. J Parkinsons Dis 2021, 11 (3), 1117-1127. Chen‐Plotkin, A. S.; Hu, W. T.; Siderowf, A.; Weintraub, D.; Goldmann Gross, R.; Hurtig, H. I.; Xie, S. X.; Arnold, S. E.; Grossman, M.; Clark, C. M., Plasma epidermal growth factor levels predict cognitive decline in Parkinson disease. Annals of neurology 2011, 69 (4), 655-663. Chiu, M.-J.; Yang, S.-Y.; Chen, T.-F.; Lin, C.-H.; Yang, F.-C.; Chen, W.-P.; Zetterberg, H.; Blennow, K., Synergistic association between plasma Aβ1–42 and p-tau in Alzheimer’s disease but not in Parkinson’s disease or frontotemporal dementia. ACS Chemical Neuroscience 2021, 12 (8), 1376-1383. Ding, J.; Zhang, J.; Wang, X.; Zhang, L.; Jiang, S.; Yuan, Y.; Li, J.; Zhu, L.; Zhang, K., Relationship between the plasma levels of neurodegenerative proteins and motor subtypes of Parkinson’s disease. Journal of neural transmission 2017, 124 , 353-360. Wong, Y. Y.; Wu, C.-Y.; Yu, D.; Kim, E.; Wong, M.; Elez, R.; Zebarth, J.; Ouk, M.; Tan, J.; Liao, J., Biofluid markers of blood-brain barrier disruption and neurodegeneration in Lewy body spectrum diseases: A systematic review and meta-analysis. Parkinsonism & Related Disorders 2022, 101 , 119-128. Batzu, L.; Rota, S.; Hye, A.; Heslegrave, A.; Trivedi, D.; Gibson, L. L.; Farrell, C.; Zinzalias, P.; Rizos, A.; Zetterberg, H., Plasma p-tau181, neurofilament light chain and association with cognition in Parkinson’s disease. npj Parkinson's Disease 2022, 8 (1), 154. Sanchez, E.; Wilkinson, T.; Coughlan, G.; Mirza, S.; Baril, A. A.; Ramirez, J.; Binns, M. A.; Black, S. E.; Borrie, M.; Dilliott, A. A.; Dixon, R. A.; Dowlatshahi, D.; Farhan, S.; Finger, E.; Fischer, C. E.; Frank, A.; Freedman, M.; Goncalves, R. A.; Grimes, D. A.; Hassan, A.; Hegele, R. A.; Kumar, S.; Lang, A. E.; Marras, C.; McLaughlin, P. M.; Orange, J. B.; Pasternak, S. H.; Pollock, B. G.; Rajji, T. K.; Roberts, A. C.; Robinson, J. F.; Rogaeva, E.; Sahlas, D. J.; Saposnik, G.; Strong, M. J.; Swartz, R. H.; Tang-Wai, D. F.; Tartaglia, M. C.; Troyer, A. K.; Kvartsberg, H.; Zetterberg, H.; Munoz, D. P.; Masellis, M., Association of plasma biomarkers with cognition, cognitive decline, and daily function across and within neurodegenerative diseases: Results from the Ontario Neurodegenerative Disease Research Initiative. Alzheimers Dement 2024, 20 (3), 1753-1770. Pedersen, C. C.; Ushakova, A.; Alves, G.; Tysnes, O.-B.; Blennow, K.; Zetterberg, H.; Maple-Grødem, J.; Lange, J., Serum neurofilament light at diagnosis: a prognostic indicator for accelerated disease progression in Parkinson’s Disease. npj Parkinson's Disease 2024, 10 (1), 162. Vijiaratnam, N.; Lawton, M.; Heslegrave, A. J.; Guo, T.; Tan, M.; Jabbari, E.; Real, R.; Woodside, J.; Grosset, K.; Chelban, V.; Athauda, D.; Girges, C.; Barker, R. A.; Hardy, J.; Wood, N.; Houlden, H.; Williams, N.; Ben-Shlomo, Y.; Zetterberg, H.; Grosset, D. G.; Foltynie, T.; Morris, H. R., Combining biomarkers for prognostic modelling of Parkinson's disease. J Neurol Neurosurg Psychiatry 2022, 93 (7), 707-15. Mattsson, N.; Cullen, N. C.; Andreasson, U.; Zetterberg, H.; Blennow, K., Association between longitudinal plasma neurofilament light and neurodegeneration in patients with Alzheimer disease. JAMA neurology 2019, 76 (7), 791-799. Yoon, S.; Kim, M.; Lee, W.-W., Long short-term memory-based deep learning models for screening Parkinson’s disease using sequential diagnostic codes. Journal of Clinical Neurology (Seoul, Korea) 2023, 19 (3), 270. Zhang, J.; Zhang, X.; Sh, Y.; Liu, B.; Hu, Z., Diagnostic AI Modeling and Pseudo Time Series Profiling of AD and PD Based on Individualized Serum Proteome Data. Frontiers in Bioinformatics 2021, 1 , 764497. Lam, S.; Arif, M.; Song, X.; Uhlén, M.; Mardinoglu, A., Machine learning analysis reveals biomarkers for the detection of neurological diseases. Frontiers in Molecular Neuroscience 2022, 15 , 889728. Vásquez-Correa, J. C.; Arias-Vergara, T.; Orozco-Arroyave, J. R.; Eskofier, B.; Klucken, J.; Nöth, E., Multimodal assessment of Parkinson's disease: a deep learning approach. IEEE journal of biomedical and health informatics 2018, 23 (4), 1618-1630. Junaid, M.; Ali, S.; Eid, F.; El-Sappagh, S.; Abuhmed, T., Explainable machine learning models based on multimodal time-series data for the early detection of Parkinson’s disease. Computer Methods and Programs in Biomedicine 2023, 234 , 107495. Mohammadi, R.; Ng, S. Y. E.; Tan, J. Y.; Ng, A. S. L.; Deng, X.; Choi, X.; Heng, D. L.; Neo, S.; Xu, Z.; Tay, K. Y.; Au, W. L.; Tan, E. K.; Tan, L. C. S.; Steyerberg, E. W.; Greene, W.; Saffari, S. E., Machine Learning for Early Detection of Cognitive Decline in Parkinson's Disease Using Multimodal Biomarker and Clinical Data. Biomedicines 2024, 12 (12). Ng, S. Y.-E.; Chia, N. S.-Y.; Abbas, M. M.; Saffari, E. S.; Choi, X.; Heng, D. L.; Xu, Z.; Tay, K.-Y.; Au, W.-L.; Tan, E.-K., Physical activity improves anxiety and apathy in early Parkinson's disease: a longitudinal follow-up study. Frontiers in Neurology 2021, 11 , 625897. Yong, A. C. W.; Tan, Y. J.; Zhao, Y.; Lu, Z.; Ng, E. Y. L.; Ng, S. Y. E.; Chia, N. S. Y.; Choi, X.; Heng, D.; Neo, S., SNCA Rep1 microsatellite length influences non-motor symptoms in early Parkinson’s disease. Aging (Albany NY) 2020, 12 (20), 20880. Stekhoven, D. J., Using the missForest package. R package 2011 , 1-11. Youden, W. J., Index for rating diagnostic tests. Cancer 1950, 3 (1), 32-35. Wang, X.; Yang, X.; He, W.; Song, X.; Zhang, G.; Niu, P.; Chen, T., The association of serum neurofilament light chains with early symptoms related to Parkinson's disease: A cross-sectional study. Journal of Affective Disorders 2023, 343 , 144-152. Welton, T.; Tan, Y. J.; Saffari, S. E.; Ng, S. Y.; Chia, N. S.; Yong, A. C.; Choi, X.; Heng, D. L.; Shih, Y.-C.; Hartono, S., Plasma neurofilament light concentration is associated with diffusion-tensor MRI-based measures of neurodegeneration in early Parkinson’s disease. Journal of Parkinson's Disease 2022, 12 (7), 2135-2146. Ng, A. S. L.; Tan, Y. J.; Yong, A. C. W.; Saffari, S. E.; Lu, Z.; Ng, E. Y.; Ng, S. Y. E.; Chia, N. S. Y.; Choi, X.; Heng, D., Utility of plasma Neurofilament light as a diagnostic and prognostic biomarker of the postural instability gait disorder motor subtype in early Parkinson’s disease. Molecular neurodegeneration 2020, 15 , 1-8. Ng, A. S. L.; Tan, Y. J.; Yong, A. C. W.; Saffari, S. E.; Lu, Z.; Ng, E. Y.; Ng, S. Y. E.; Chia, N. S. Y.; Choi, X.; Heng, D.; Neo, S.; Xu, Z.; Keong, N. C. H.; Tay, K. Y.; Au, W. L.; Tan, L. C. S.; Tan, E. K., Utility of plasma Neurofilament light as a diagnostic and prognostic biomarker of the postural instability gait disorder motor subtype in early Parkinson's disease. Mol Neurodegener 2020, 15 (1), 33. Ygland Rödström, E.; Mattsson-Carlgren, N.; Janelidze, S.; Hansson, O.; Puschmann, A., Serum neurofilament light chain as a marker of progression in Parkinson’s disease: Long-term observation and implications of clinical subtypes. Journal of Parkinson's Disease 2022, 12 (2), 571-584. Halloway, S.; Desai, P.; Beck, T.; Aggarwal, N.; Agarwal, P.; Evans, D.; Rajan, K. B., Association of Neurofilament Light With the Development and Severity of Parkinson Disease. Neurology 2022, 98 (22), e2185-e2193. Ou, R.; Liu, K.; Lin, J.; Yang, T.; Xiao, Y.; Wei, Q.; Hou, Y.; Li, C.; Zhang, L.; Jiang, Z.; Zhao, B.; Chen, X.; Song, W.; Wu, Y.; Shang, H., Relationship between plasma NFL and disease progression in Parkinson's disease: a prospective cohort study. J Neurol 2024, 271 (4), 1837-1843. Mollenhauer, B.; Dakna, M.; Kruse, N.; Galasko, D.; Foroud, T.; Zetterberg, H.; Schade, S.; Gera, R. G.; Wang, W.; Gao, F., Validation of serum neurofilament light chain as a biomarker of Parkinson's disease progression. Movement Disorders 2020, 35 (11), 1999-2008. Kwon, K.-Y.; Pyo, S. J.; Lee, H. M.; Seo, W.-K.; Koh, S.-B., Cognition and visit-to-visit variability of blood pressure and heart rate in de novo patients with Parkinson’s disease. Journal of Movement Disorders 2016, 9 (3), 144. Doiron, M.; Langlois, M.; Dupré, N.; Simard, M., The influence of vascular risk factors on cognitive function in early Parkinson's disease. International Journal of Geriatric Psychiatry 2018, 33 (2), 288-297. Xiao, Y.; Yang, T.; Zhang, L.; Wei, Q.; Ou, R.; Hou, Y.; Liu, K.; Lin, J.; Jiang, Q.; Shang, H., Association between the blood pressure variability and cognitive decline in Parkinson's disease. Brain Behav 2023, 13 (12), e3319. Irwin, D. J.; Grossman, M.; Weintraub, D.; Hurtig, H. I.; Duda, J. E.; Xie, S. X.; Lee, E. B.; Van Deerlin, V. M.; Lopez, O. L.; Kofler, J. K., Neuropathological and genetic correlates of survival and dementia onset in synucleinopathies: a retrospective analysis. The Lancet Neurology 2017, 16 (1), 55-65. Robinson, J. L.; Lee, E. B.; Xie, S. X.; Rennert, L.; Suh, E.; Bredenberg, C.; Caswell, C.; Van Deerlin, V. M.; Yan, N.; Yousef, A., Neurodegenerative disease concomitant proteinopathies are prevalent, age-related and APOE4-associated. Brain 2018, 141 (7), 2181-2193. Gomperts, S. N.; Locascio, J. J.; Makaretz, S. J.; Schultz, A.; Caso, C.; Vasdev, N.; Sperling, R.; Growdon, J. H.; Dickerson, B. C.; Johnson, K., Tau positron emission tomographic imaging in the Lewy body diseases. JAMA neurology 2016, 73 (11), 1334-1341. Compta, Y.; Parkkinen, L.; O'Sullivan, S. S.; Vandrovcova, J.; Holton, J. L.; Collins, C.; Lashley, T.; Kallis, C.; Williams, D. R.; de Silva, R., Lewy-and Alzheimer-type pathologies in Parkinson's disease dementia: which is more important? Brain 2011, 134 (5), 1493-1505. Jellinger, K. A., Dementia with Lewy bodies and Parkinson’s disease-dementia: current concepts and controversies. Journal of neural transmission 2018, 125 , 615-650. McKeith, I. G.; Boeve, B. F.; Dickson, D. W.; Halliday, G.; Taylor, J.-P.; Weintraub, D.; Aarsland, D.; Galvin, J.; Attems, J.; Ballard, C. G., Diagnosis and management of dementia with Lewy bodies: Fourth consensus report of the DLB Consortium. Neurology 2017, 89 (1), 88-100. Espay, A. J.; Schwarzschild, M. A.; Tanner, C. M.; Fernandez, H. H.; Simon, D. K.; Leverenz, J. B.; Merola, A.; Chen‐Plotkin, A.; Brundin, P.; Kauffman, M. A., Biomarker‐driven phenotyping in Parkinson's disease: a translational missing link in disease‐modifying clinical trials. Movement Disorders 2017, 32 (3), 319-324. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 25 Feb, 2026 Read the published version in npj Parkinson's Disease → Version 1 posted Editorial decision: Revision requested 15 Aug, 2025 Reviews received at journal 14 Aug, 2025 Reviews received at journal 01 Aug, 2025 Reviewers agreed at journal 22 Jul, 2025 Reviewers agreed at journal 22 Jul, 2025 Reviewers agreed at journal 22 Jul, 2025 Reviewers invited by journal 22 Jul, 2025 Editor assigned by journal 15 Jul, 2025 Submission checks completed at journal 15 Jul, 2025 First submitted to journal 12 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7107548","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":489323670,"identity":"d82bf726-8c15-4b31-8079-c39ba22f6ffd","order_by":0,"name":"Raziyeh Mohammadi","email":"","orcid":"","institution":"National University of Singapore","correspondingAuthor":false,"prefix":"","firstName":"Raziyeh","middleName":"","lastName":"Mohammadi","suffix":""},{"id":489323671,"identity":"b39f8e3c-d210-48f0-ab28-9816f1d4a5b7","order_by":1,"name":"Samuel Y. E. Ng","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Samuel","middleName":"Y. E.","lastName":"Ng","suffix":""},{"id":489323672,"identity":"31343396-39bc-4466-b67c-d53e5b6d4bca","order_by":2,"name":"Jayne Y. Tan","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Jayne","middleName":"Y.","lastName":"Tan","suffix":""},{"id":489323673,"identity":"15eb4303-b4ea-4549-8eb2-400a2c909496","order_by":3,"name":"Adeline S. L. Ng","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Adeline","middleName":"S. L.","lastName":"Ng","suffix":""},{"id":489323674,"identity":"461e9a98-20ea-4131-bb6e-bf49acf6377c","order_by":4,"name":"Xiao Deng","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Xiao","middleName":"","lastName":"Deng","suffix":""},{"id":489323675,"identity":"bbde8f05-771d-46c0-8070-98f566ba6ff3","order_by":5,"name":"Xinyi Choi","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Xinyi","middleName":"","lastName":"Choi","suffix":""},{"id":489323676,"identity":"becd52c2-7e0f-4d33-a5b4-70c13cc8bbd6","order_by":6,"name":"Dede L. Heng","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Dede","middleName":"L.","lastName":"Heng","suffix":""},{"id":489323677,"identity":"c7fe363e-bcf0-4e98-a687-92534ef79e21","order_by":7,"name":"Shermyn Neo","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Shermyn","middleName":"","lastName":"Neo","suffix":""},{"id":489323678,"identity":"e4768990-5966-40b3-be53-99f9bb397a77","order_by":8,"name":"Zheyu Xu","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Zheyu","middleName":"","lastName":"Xu","suffix":""},{"id":489323679,"identity":"0ce0221e-09ab-4eac-8d71-7b54100793bf","order_by":9,"name":"Kay-Yaw Tay","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Kay-Yaw","middleName":"","lastName":"Tay","suffix":""},{"id":489323680,"identity":"0c70c064-4d41-4671-bf9b-8b9305a0d993","order_by":10,"name":"Wing-Lok Au","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Wing-Lok","middleName":"","lastName":"Au","suffix":""},{"id":489323681,"identity":"2cbe45a3-e8a3-400d-a3ee-153ae8f8da4b","order_by":11,"name":"Eng-King Tan","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Eng-King","middleName":"","lastName":"Tan","suffix":""},{"id":489323682,"identity":"b57343c2-d55a-494e-ad45-6816abe4ad0b","order_by":12,"name":"Louis C. S. Tan","email":"","orcid":"","institution":"National Neuroscience Institute","correspondingAuthor":false,"prefix":"","firstName":"Louis","middleName":"C. S.","lastName":"Tan","suffix":""},{"id":489323683,"identity":"b630af25-97a0-42b7-acfa-52272530c22c","order_by":13,"name":"William Greene","email":"","orcid":"","institution":"Stern School of Business, New York University","correspondingAuthor":false,"prefix":"","firstName":"William","middleName":"","lastName":"Greene","suffix":""},{"id":489323684,"identity":"e3a84880-d8b7-4bbd-8e92-63e0d01c52d3","order_by":14,"name":"Maria Liakata","email":"","orcid":"","institution":"Queen Mary University of London","correspondingAuthor":false,"prefix":"","firstName":"Maria","middleName":"","lastName":"Liakata","suffix":""},{"id":489323685,"identity":"58e29e54-026c-40e5-9474-4c69170ebeb3","order_by":15,"name":"Seyed Ehsan Saffari","email":"data:image/png;base64,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","orcid":"","institution":"National University of Singapore","correspondingAuthor":true,"prefix":"","firstName":"Seyed","middleName":"Ehsan","lastName":"Saffari","suffix":""}],"badges":[],"createdAt":"2025-07-12 10:38:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7107548/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7107548/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41531-026-01298-8","type":"published","date":"2026-02-25T15:58:13+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":87701710,"identity":"7f6786d6-7514-4529-883c-f9cf33d4627e","added_by":"auto","created_at":"2025-07-28 07:25:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":139730,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature importance rankings for predicting cognitive decline in early PD.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTop row, left: The Random Forest model uses mean decrease in Gini index to measure how each feature contributes to reducing impurity in decision tree splits. This metric reflects how often and effectively a feature is used for classification across all trees in the forest. Top row, right: The ShapleyVIC method combines Shapley values with Variable Inclusion Confidence, providing statistical confidence about feature importance. Variables with 95% prediction intervals above zero are considered reliably important predictors. Bottom row: XGBoost's gain-based importance scores quantify each feature's contribution to model accuracy by measuring the improvement in prediction accuracy when that feature is used for tree splits.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7107548/v1/255ab150cdaa2221187e4329.png"},{"id":87702356,"identity":"637ab5df-5cab-4cac-8461-28651e7422a0","added_by":"auto","created_at":"2025-07-28 07:33:01","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":91698,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCalibration plots for five predictive models using Model 4. Each panel shows a calibration plot for a different model: Random Forest (RF), Neural Network (NN), K-Nearest Neighbors (KNN), XGBoost, and LightGBM. The solid black line connects the observed event rates for each bin of predicted probabilities. Black circles represent the actual (observed) event rate within each bin of predicted probabilities, plotted against the midpoint of each bin. The blue shaded area indicates the 99% confidence interval around the observed event rates. The dashed blue diagonal line represents perfect calibration, where predicted probabilities match actual event rates.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7107548/v1/c433124ec15d8d56e45f3dbc.png"},{"id":103765636,"identity":"58b13ae8-c057-4541-baf3-aff114266aea","added_by":"auto","created_at":"2026-03-02 16:06:10","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1825954,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7107548/v1/5a528f17-d456-4870-acdb-712607dbb104.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine Learning Integration of Serial Blood Biomarkers Enhances Cognitive Decline Prediction in Early Parkinson's Disease","fulltext":[{"header":"Introduction","content":"\u003cp\u003eParkinson disease (PD) is the second most common neurodegenerative disorder, affecting over 10\u0026nbsp;million people globally. With aging populations, its prevalence is expected to double by 2040, increasing burdens on healthcare systems and caregivers. Although PD is primarily defined by motor symptoms such as bradykinesia, rigidity, and tremor, non-motor features—particularly cognitive impairment—contribute significantly to morbidity [\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e–\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Cognitive decline affects up to 80% of patients, with many progressing from mild cognitive impairment (PD-MCI) to Parkinson’s disease dementia (PDD) within a few years of diagnosis [\u003cspan additionalcitationids=\"CR6 CR7 CR8 CR9\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e–\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Cognitive deficits in PD have serious clinical and societal impacts, including reduced quality of life, loss of independence, and higher rates of institutionalization. Healthcare costs for patients with PDD are twice those without dementia, and caregivers face increased psychological and financial strain. Predicting cognitive decline remains challenging due to heterogeneous disease progression. Early identification of at-risk patients is crucial for timely intervention, counseling, and improving clinical trial design [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan additionalcitationids=\"CR12 CR13\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e–\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eCurrent prognostic tools poorly predict cognitive outcomes in PD. Clinical scales like the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) focus on motor symptoms and offer limited insight into cognitive decline. While Cerebrospinal fluid (CSF) biomarkers and neuroimaging provide some prognostic value, they are invasive, costly, and impractical for routine use [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. In contrast, blood-based biomarkers are minimally invasive, cost-effective, and feasible for repeated assessments [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Neurofilament light chain (NfL), a marker of axonal injury, correlates strongly with CSF levels and is linked to disease severity and cognitive decline in PD [\u003cspan additionalcitationids=\"CR20 CR21 CR22\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e–\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Similarly, total tau (t-tau) and phosphorylated tau181 (p-tau181) have been associated with cognitive dysfunction [\u003cspan additionalcitationids=\"CR25\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e–\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Prior studies have reported elevated plasma NfL and t-tau in individuals with PD who exhibit early cognitive deficits, and combinations of plasma α-synuclein, p-tau181, and Aβ-40 have been proposed as potential biomarkers [\u003cspan additionalcitationids=\"CR27 CR28 CR29 CR30 CR31 CR32 CR33\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e–\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. However, the vast majority of these investigations rely on cross-sectional data, limiting their predictive power. Longitudinal biomarker trajectories, such as changes in NfL and t-tau, may better reflect disease progression [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e], yet few studies have evaluated their predictive utility in early PD or combined them with clinical data to improve prognostic accuracy [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eAdvances in machine learning (ML) have transformed neurodegeneration research by enabling the analysis of complex, high-dimensional data [\u003cspan additionalcitationids=\"CR39\" citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e–\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. ML can detect non-linear patterns and interactions often missed by traditional methods[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Prior work using data from the Parkinson’s Progression Markers Initiative (PPMI) has shown that ML models can achieve high accuracy in predicting disease progression by incorporating multimodal data, including motor scores, biospecimens, and patient-reported outcomes [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. However, the utility of ML for predicting cognitive decline in early PD using time-varying blood biomarkers remains underexplored.\u003c/p\u003e\u003cp\u003eBuilding on our prior work predicting cognitive decline in early PD using baseline features (AUC = 0.93) [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e], we now examine whether adding longitudinal plasma NfL and t-tau measurements improves prediction. Using data from the Early Parkinson’s Disease Longitudinal Singapore (PALS) cohort, we apply multiple ML algorithms to model cognitive decline over five years. Our objectives are to: (1) assess the added value of longitudinal biomarkers, (2) identify key predictors using advanced feature selection, and (3) develop a clinically relevant risk prediction tool.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cb\u003eStudy Design, Participants, and Variables\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThis prospective observational study used data from the PALS cohort, which tracks clinical, cognitive, and biological changes in early PD. Participants were enrolled from specialist clinics in Singapore (2014–2018). Inclusion criteria were PD diagnosis within 1 year, symptom onset within 2 years, and meeting the National Institute of Neurological Disorders and Stroke criteria. Participants required at least six years of education. Exclusion criteria included comorbidities likely to confound outcomes or hinder follow-up (e.g., malignancy, unstable cardiovascular disease) [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. Of 214 participants recruited, 193 completed baseline assessments and had at least three years of cognitive follow-up, comprising the final analytic cohort. Informed consent was obtained; the study was approved by the Singapore Health Services Centralized Institutional Review Board.\u003c/p\u003e\u003cp\u003e Participants underwent detailed annual evaluations over five years, including clinical, cognitive, and laboratory assessments. Cognitive performance was measured using the Montreal Cognitive Assessment (MoCA). Venous blood samples were collected at baseline, year 3, and year 5. Primary biomarkers—NfL and t-tau—were quantified using ultra-sensitive single-molecule array (Simoa) assays. For these biomarkers, four summary features (minimum, maximum, mean, and standard deviation) were computed across time points to capture both levels and variability. Additional baseline biomarkers included suppression of tumorigenicity 2 (ST2), apolipoprotein E (APOE) genotype, p-tau181, and alpha-synuclein gene promoter repeat length (REP1). The primary outcome was cognitive decline, defined as either a ≥ 1-point annual decrease in MoCA score or a ≥ 1-point drop sustained across two consecutive years during the five-year follow-up. This threshold was selected to reflect subtle but clinically meaningful cognitive changes. Baseline demographic characteristics included age, sex, body mass index (BMI), years of education, smoking history, alcohol use, and consumption of coffee and tea. Clinical features comprised Hoehn and Yahr stage (HY), baseline MoCA score, MDS-UPDRS Part III score, systolic and diastolic blood pressure (SBP and DBP) measured in both lying and standing positions, and comorbidities including hypertension, diabetes mellitus, hypertension and hyperlipidemia.\u003c/p\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eMissing data were addressed using a random forest (RF)–based imputation method [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. All continuous variables were examined for clinical interpretability, and when appropriate, were dichotomized using clinically relevant thresholds. In the absence of established cutoffs, the Youden index was used to derive optimal cut-points from ROC curves to maximize sensitivity and specificity [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. Descriptive statistics were used to summarize all variables. Continuous variables were presented as mean (standard deviation) or median (interquartile range), and categorical variables as counts and percentages. Associations between individual predictors and cognitive decline were assessed using univariate logistic regression, reporting odds ratios (ORs) with 95% confidence intervals (CIs).\u003c/p\u003e\u003cp\u003eTo identify the most informative predictors, we applied three complementary variable selection methods. First, RF importance scores were used to rank variables by mean decrease in Gini impurity, retaining only those above the average score. Second, the Shapley Variable Importance Cloud (SHAPvIC) method quantified each variable’s contribution using SHAP values; variables with 95% prediction intervals entirely above zero were selected. Third, we applied gain-based feature importance from Extreme Gradient Boosting (XGBoost), keeping variables with scores above the mean. This triangulated approach enhanced both statistical robustness and consistency across methods. For predictive modeling, we trained five supervised ML algorithms: RF, neural network (NN), k-nearest neighbors (KNN), XGBoost, and Light Gradient Boosting Machine (LightGBM). Each algorithm was evaluated across seven different feature sets: time-varying biomarkers alone; baseline variables alone; a combination of both; and subsets derived from each of the three variable selection strategies (including a reduced version of SHAPvIC for parsimony). The dataset was randomly split (80:20) into training and test sets, stratified by cognitive decline status. Hyperparameters were optimized via five-fold cross-validation on the training set. For RF, we tuned the number of trees and number of predictors considered at each split. For XGBoost and LightGBM, we optimized learning rate (0.01–0.1), number of boosting rounds (50–500), maximum tree depth (3–7), and L1/L2 regularization terms (range: 0–1). NN models varied in hidden layer size (1–5) and applied weight decay (0–0.1), while the optimal number of neighbors (k = 1–10) was selected for KNN models. Model performance was assessed on the held-out test set using multiple metrics. Discrimination was evaluated via AUC with 95% CIs from 1,000 bootstrap resamples. Additional metrics included sensitivity, specificity, accuracy, precision, recall, F1 score, and Brier score. The F1 score, the harmonic mean of precision and recall, balances false positives and negatives. The Brier score quantifies the accuracy of probabilistic predictions. Sensitivity and specificity were calculated at the optimal threshold via Youden’s index. Calibration was assessed using decile-based plots comparing observed and predicted event rates across 10 bins. All analyses were performed using R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was set at a two-sided \u003cem\u003eP\u003c/em\u003e Value \u0026lt; 0.05.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eA total of 193 participants with early PD were included. At baseline, the mean age was 63.8 years (SD\u0026thinsp;=\u0026thinsp;8.9), 58% were male, and the median education duration was 11 years (IQR\u0026thinsp;=\u0026thinsp;8\u0026ndash;15). The mean MoCA score was 25.4 (SD\u0026thinsp;=\u0026thinsp;2.8). Disease severity was mild to moderate: 62% in HY stage 1, 30% in stage 2, and 8% in stage 3. Over five years, 44 participants (22.8%) experienced cognitive decline (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Higher education (\u0026ge;\u0026thinsp;10 years) was protective (OR\u0026thinsp;=\u0026thinsp;0.4; 95% CI: 0.2\u0026ndash;0.7; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.0006), while elevated blood pressure increased risk: lying DBP (OR\u0026thinsp;=\u0026thinsp;2.6; 95% CI: 1.3\u0026ndash;5.2; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.009), standing DBP (OR\u0026thinsp;=\u0026thinsp;2.1; 95% CI: 1.1\u0026ndash;4.3; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.045), and lying SBP (OR\u0026thinsp;=\u0026thinsp;2.1; 95% CI: 1.1\u0026ndash;4.3; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.045) (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSummary statistics of demographic, clinical assessments, blood biomarkers, and their associations with progression outcomes using univariate logistic regression.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNo Progression\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eProgression\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eOR\u003c/p\u003e\u003cp\u003e(95% CIs)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e Value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;193\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;149\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;44\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e\u003cp\u003eDemographic Characteristics\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMale Gender\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e112 (58.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e85 (57.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e27 (61.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.2 (0.6, 2.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.737\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSmoker\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e56 (29.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e43 (28.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e13 (29.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (0.5, 2.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.000\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYears of education (\u0026ge;\u0026thinsp;10 years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e135 (69.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e112 (75.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23 (52.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.4 (0.2, 0.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.006\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTea drinking\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e180 (93.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e138 (92.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e42 (95.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.6 (0.4, 11.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.736\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCoffee drinking\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e175 (90.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e135 (90.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e40 (90.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (0.3, 3.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.000\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAlcohol drinking\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e125 (64.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e101 (67.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24 (54.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.6 (0.3, 1.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.151\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBMI (\u0026gt;\u0026thinsp;25 kg/m2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e64 (33.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e48 (32.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e16 (36.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.2 (0.6, 2.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.740\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge (\u0026gt;\u0026thinsp;65 years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e97 (50.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e72 (48.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e25 (56.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.4 (0.7, 2.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.413\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e\u003cp\u003eClinical Assessments\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLying SBP (\u0026ge;\u0026thinsp;140 mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e95 (49.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e67 (45.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28 (63.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2.1 (1.1, 4.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.045\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLying DBP (\u0026ge;\u0026thinsp;80 mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e67 (34.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e44 (29.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23 (52.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2.6 (1.3, 5.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.009\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStanding SBP (\u0026ge;\u0026thinsp;140 mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e85 (44.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e63 (42.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22 (50.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.4 (0.7, 2.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.463\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStanding DBP (\u0026ge;\u0026thinsp;80 mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e95 (49.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e67 (45.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28 (63.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2.1 (1.1, 4.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.045\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiabetes mellitus\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e31 (16.1%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e25 (16.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6 (13.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.8 (0.3, 2.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.791\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e88 (45.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e68 (45.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e20 (45.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (0.5, 2.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.000\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHyperlipidemia\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e92 (47.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e73 (49.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e19 (43.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.8 (0.4, 1.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.613\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMoCA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e26 [23.0, 28.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e26 [23.0, 28.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e26 [23.0, 28.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (0.9, 1.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.771\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTotal motor score\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20.0 [15.0; 26.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e19.0 [15.0; 26.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22.0 [17.0; 29.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (1.0, 1.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.062\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHY\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.00 [1.0; 3.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.00 [1.50; 2.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.00 [2.00; 2.0]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2.0 (0.8, 4.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.112\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e\u003cp\u003eBlood Biomarkers\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAPOE4 (Non-carriers)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e153 (79.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e120 (80.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e33 (75.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.7 (0.3, 1.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.559\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRep 1 (Short)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e88 (45.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e66 (44.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22 (50.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.3 (0.6, 2.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.620\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eST2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11,596 [8747; 14,830]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11,534 [8,402; 14,854]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12611 [9,433; 14,813]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (1.0, 1.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.375\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNfL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e13.7 [10.1; 18.9]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e13.9 [10.2; 18.7]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e13.3 [9.9; 21.7]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (1.0, 1.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.702\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003et-tau\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.17 [0.9; 1.5]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.1 [0.9; 1.6]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.3 [0.9; 1.5]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.3 (0.9, 1.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.350\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ep-tau181\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20.3 [15.7; 24.8]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20.5 [15.4; 24.3]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e20.1 [15.8; 28.9]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.0 (1.0, 1.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.666\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eData are expressed as frequency (%) or median (quartile); \u003cem\u003eP\u003c/em\u003e values are from univariate logistic regression models assessing the association of each variable with cognitive decline progression. Abbreviations: N: number, OR: odds ratio, CIs: confidence intervals, BMI: body mass index, SBP: systolic blood pressure, DBP: diastolic blood pressure, MoCA: Montreal Cognitive Assessment, HY: Hoehn and Yahr scale, APOE: apolipoprotein E, REP1: alpha-synuclein gene promoter, ST2: suppression of tumorigenicity 2, NfL: neurofilament light chain, t-tau: total tau, p-tau181: phosphorylated tau at threonine 181.\u003c/p\u003e\u003cp\u003eTo identify the most predictive variables, thirty candidate features were assessed using three feature selection methods. The RF approach identified twelve variables, including years of education, lying and standing DBP, and four summary statistics (minimum, maximum, mean, and standard deviation) for both NfL and t-tau, along with baseline p-tau181 levels (Fig.\u0026nbsp;1, top row, left). The ShapleyVIC method selected fourteen predictors, which included t-tau (minimum, maximum, mean, and standard deviation), NfL (minimum and mean), as well as baseline MoCA score, hyperlipidemia, coffee and alcohol consumption, lying and standing DBP, p-tau181, and gender (Fig.\u0026nbsp;1, top row, right). The XGBoost-based method identified eleven key features, including the full set of summary metrics for both NfL and t-tau, lying and standing DBP, and years of education (Fig.\u0026nbsp;1, bottom row). Notably, eight variables\u0026mdash;minimum, maximum, mean, and standard deviation of t-tau; minimum and mean NfL; and lying and standing DBP\u0026mdash;were consistently selected across all three methods.\u003c/p\u003e\u003cp\u003eML performance varied across algorithms and feature sets (Table\u0026nbsp;2). Models using longitudinal biomarker summaries (Model 1) consistently outperformed those using only baseline variables (Model 2), with AUCs ranging from 0.640 to 0.763 vs 0.465 to 0.560, respectively.\u003c/p\u003e\u003cp\u003eThe XGBoost model using RF-selected variables (Model 4) achieved the highest AUC (0.806; 95% CI, 0.663\u0026ndash;0.949) and the best overall performance, with accuracy of 0.730, sensitivity of 0.875, specificity of 0.690, F1 score of 0.583, and a Brier score of 0.140. The LightGBM model using time-varying biomarker summaries (Model 1) performed well (AUC\u0026thinsp;=\u0026thinsp;0.763; 95% CI, 0.598\u0026ndash;0.928), as did both LightGBM and XGBoost models using SHAPvIC-selected variables (Model 5; AUC\u0026thinsp;=\u0026thinsp;0.698 for both). These models showed comparable sensitivity and specificity to the top-performing XGBoost model, though with slightly lower precision and calibration. Notably, the reduced SHAPvIC subset (Model 6), comprising six variables, achieved the highest specificity (0.862; 95% CI, 0.683\u0026ndash;0.961) with XGBoost, highlighting the utility of parsimonious models in clinical prediction.\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e\u003cb\u003eTable\u0026nbsp;2. Performance of ML models across different feature sets.\u003c/b\u003e\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"684\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel 1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel 2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel 3\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel 4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel 5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel 6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel 7\u003c/strong\u003e\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.763\u003c/p\u003e\n \u003cp\u003e(0.613, 0.912)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.478\u003c/p\u003e\n \u003cp\u003e(0.201,0.755)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.6056\u003c/p\u003e\n \u003cp\u003e(0.395, 0.817)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.6853\u003c/p\u003e\n \u003cp\u003e(0.499, 0.871)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.733\u003c/p\u003e\n \u003cp\u003e(0.520, 0.946)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.5345\u003c/p\u003e\n \u003cp\u003e(0.299, 0.770)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.580\u003c/p\u003e\n \u003cp\u003e(0.351, 0.808)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003cp\u003e(0.473, 0.997)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003cp\u003e(0.157, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003cp\u003e(0.157, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.655\u003c/p\u003e\n \u003cp\u003e(0.457, 0.821)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003cp\u003e(0.565, 0.897)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.483\u003c/p\u003e\n \u003cp\u003e(0.294, 0.675)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.621\u003c/p\u003e\n \u003cp\u003e(0.423, 0.793)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003cp\u003e(0.492, 0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.621\u003c/p\u003e\n \u003cp\u003e(0.423, 0.793)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.483\u003c/p\u003e\n \u003cp\u003e(0.294, 0.675)\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003cp\u003e(0.530, 0.841)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003cp\u003e(0.530, 0.841)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.540\u003c/p\u003e\n \u003cp\u003e(0.369, 0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.649\u003c/p\u003e\n \u003cp\u003e(0.475, 0.798)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003cp\u003e(0.530, 0.841)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.595\u003c/p\u003e\n \u003cp\u003e(0.421, 0.752)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.540\u003c/p\u003e\n \u003cp\u003e(0.369, 0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.4118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.364\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.286\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.353\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.286\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eF1 Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.560\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.421\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.414\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.480\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.522\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.348\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.414\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eLog Loss\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eNE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.529\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.514\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.478\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.567\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eBrier Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.157\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.2010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.154\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.180\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.711\u003c/p\u003e\n \u003cp\u003e(0.542, 0.880)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.534\u003c/p\u003e\n \u003cp\u003e(0.267, 0.802)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.685\u003c/p\u003e\n \u003cp\u003e(0.497, 0.874)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.668\u003c/p\u003e\n \u003cp\u003e(0.485, 0.852)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.5948\u003c/p\u003e\n \u003cp\u003e(0.386, 0.804)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003cp\u003e(0.370, 0.854)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.655\u003c/p\u003e\n \u003cp\u003e(0.468, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003cp\u003e(0.473, 0.997)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003cp\u003e(0.473, 0.997)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.6250\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.7500\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.8750\u003c/p\u003e\n \u003cp\u003e(0.4735, 0.9968)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.586\u003c/p\u003e\n \u003cp\u003e(0.389, 0.765)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.517\u003c/p\u003e\n \u003cp\u003e(0.325, 0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003cp\u003e(0.565, 0.897)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.517\u003c/p\u003e\n \u003cp\u003e(0.325, 0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.6897\u003c/p\u003e\n \u003cp\u003e(0.492, 0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.6552\u003c/p\u003e\n \u003cp\u003e(0.457, 0.821)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.483\u003c/p\u003e\n \u003cp\u003e(0.294, 0.675)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.649\u003c/p\u003e\n \u003cp\u003e(0.475, 0.798)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.5405\u003c/p\u003e\n \u003cp\u003e(0.369, 0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.730\u003c/p\u003e\n \u003cp\u003e(0.559, 0.862)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.595\u003c/p\u003e\n \u003cp\u003e(0.421, 0.753)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003cp\u003e(0.502, 0.820)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003cp\u003e(0.502, 0.820)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.568\u003c/p\u003e\n \u003cp\u003e(0.395, 0.729)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.368\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.263\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.417\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.357\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.375\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.318\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eF1 Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.518\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.370\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.483\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.454\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.467\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eLog Loss\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1.046\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.775\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.634\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.809\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.529\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.556\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eBrier Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.159\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.232\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.239\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.214\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.254\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.172\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.194\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKNN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.640\u003c/p\u003e\n \u003cp\u003e(0.440, 0.840)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.560\u003c/p\u003e\n \u003cp\u003e(0.331, 0.789)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.619\u003c/p\u003e\n \u003cp\u003e(0.408, 0.829)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.640\u003c/p\u003e\n \u003cp\u003e(0.436, 0.844)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.534\u003c/p\u003e\n \u003cp\u003e(0.338-0.731)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.603\u003c/p\u003e\n \u003cp\u003e(0.433, 0.774)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.608\u003c/p\u003e\n \u003cp\u003e(0.421, 0.794)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003cp\u003e(0.157, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003cp\u003e(0.157, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.375\u003c/p\u003e\n \u003cp\u003e(0.085, 0.755)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.655\u003c/p\u003e\n \u003cp\u003e(0.457, 0.821)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.448\u003c/p\u003e\n \u003cp\u003e(0.264, 0.643)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.517\u003c/p\u003e\n \u003cp\u003e(0.325, 0.706)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.655\u003c/p\u003e\n \u003cp\u003e(0.457, 0.821)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.517\u003c/p\u003e\n \u003cp\u003e(0.325, 0.706)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.621\u003c/p\u003e\n \u003cp\u003e(0.423, 0.793)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.724\u003c/p\u003e\n \u003cp\u003e(0.528, 0.873)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.6486\u003c/p\u003e\n \u003cp\u003e(0.475, 0.798)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.487\u003c/p\u003e\n \u003cp\u003e(0.319, 0.656)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.568\u003c/p\u003e\n \u003cp\u003e(0.395, 0.729)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.649\u003c/p\u003e\n \u003cp\u003e(0.475, 0.798)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.513\u003c/p\u003e\n \u003cp\u003e(0.344, 0.681)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.595\u003c/p\u003e\n \u003cp\u003e(0.42, 0.752)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.649\u003c/p\u003e\n \u003cp\u003e(0.475, 0.798)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.238\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.300\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.222\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.273\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eF1 Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.435\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.345\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.429\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.435\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.308\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.348\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.316\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eLog Loss\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eInf\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.547\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.614\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.506\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eInf\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.512\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.502\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eBrier Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.351\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.177\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.206\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.167\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.173\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.168\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003cp\u003e(0.546, 0.928)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.517\u003c/p\u003e\n \u003cp\u003e(0.238, 0.797)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.595\u003c/p\u003e\n \u003cp\u003e(0.412, 0.778)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.806\u003c/p\u003e\n \u003cp\u003e(0.663, 0.949)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.698\u003c/p\u003e\n \u003cp\u003e(0.459, 0.938)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.647\u003c/p\u003e\n \u003cp\u003e(0.411, 0.882)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.659\u003c/p\u003e\n \u003cp\u003e(0.478, 0.841)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003cp\u003e(0.473, 0.997)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003cp\u003e(0.473, 0.997)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003cp\u003e(0.157, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003cp\u003e(0.473, 0.997)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.621\u003c/p\u003e\n \u003cp\u003e(0.423, 0.793)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003cp\u003e(0.492, 0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.517\u003c/p\u003e\n \u003cp\u003e(0.325, 0.706)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003cp\u003e(0.492, 0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.828\u003c/p\u003e\n \u003cp\u003e(0.642, 0.942)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.862\u003c/p\u003e\n \u003cp\u003e(0.683, 0.961)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.552\u003c/p\u003e\n \u003cp\u003e(0.357, 0.736)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.649\u003c/p\u003e\n \u003cp\u003e(0.475, 0.798)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003cp\u003e(0.502, 0.820)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.595\u003c/p\u003e\n \u003cp\u003e(0.421, 0.752)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.730\u003c/p\u003e\n \u003cp\u003e(0.550, 0.862)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.784\u003c/p\u003e\n \u003cp\u003e(0.618, 0.902)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.784\u003c/p\u003e\n \u003cp\u003e(0.618, 0.902)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.622\u003c/p\u003e\n \u003cp\u003e(0.448, 0.775)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.353\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.357\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.437\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.350\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eF1 Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.480\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.454\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.483\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.583\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.556\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eLog Loss\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.451\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.9159\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.565\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.417\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.474\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.496\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.548\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eBrier Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.146\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.230\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.140\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.152\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.160\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.181\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLightGBM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.763\u003c/p\u003e\n \u003cp\u003e(0.598, 0.928)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.465\u003c/p\u003e\n \u003cp\u003e(0.191, 0.740)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.681\u003c/p\u003e\n \u003cp\u003e(0.490, 0.872)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.728\u003c/p\u003e\n \u003cp\u003e(0.535, 0.922)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.698\u003c/p\u003e\n \u003cp\u003e(0.476, 0.920)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.465\u003c/p\u003e\n \u003cp\u003e(0.228, 0.703)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003cp\u003e(0.501, 0.878)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003cp\u003e(0.157, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003cp\u003e(0.349, 0.968)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.625\u003c/p\u003e\n \u003cp\u003e(0.245, 0.915)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003cp\u003e(0.157, 0.843)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003cp\u003e(0.473, 0.997)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003cp\u003e(0.565, 0.897)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003cp\u003e(0.492, 0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.724\u003c/p\u003e\n \u003cp\u003e(0.528, 0.873)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003cp\u003e(0.492, 0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.828\u003c/p\u003e\n \u003cp\u003e(0.642, 0.942)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003cp\u003e(0.492, 0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.552\u003c/p\u003e\n \u003cp\u003e(0.357, 0.736)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.757\u003c/p\u003e\n \u003cp\u003e(0.588, 0.882)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.649\u003c/p\u003e\n \u003cp\u003e(0.475, 0.797)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003cp\u003e(0.530, 0.841)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003cp\u003e(0.530, 0.841)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.784\u003c/p\u003e\n \u003cp\u003e(0.618, 0.902)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.6486\u003c/p\u003e\n \u003cp\u003e(0.475, 0.798)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.622\u003c/p\u003e\n \u003cp\u003e(0.448, 0.775)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.461\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.308\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.385\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.308\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.350\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eF1 Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.381\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.476\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.522\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.556\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.381\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eLog Loss\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.567\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.908\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.5555\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.537\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.488\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.526\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.624\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\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 valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eBrier Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.234\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.173\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e0.158\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.203\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 \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAbbreviations: NE: Not estimable due to infinite log loss arising from zero predicted probabilities. Evaluation was conducted using the test dataset from an 80/20 train-test split, with hyperparameters optimized via 10-fold cross-validation on the training set. Model 1: Descriptive statistics (mean, SD, Min, Max) of eight time-varying biomarkers. Model 2: 24 baseline variables. Model 3: Combined time-varying biomarkers and baseline variables (30 features). Model 4: Top 12 variables selected by RF (years of education; standing and lying DBP; four summary statistics (minimum, maximum, mean, Standard deviation) of both NfL and t-tau; and pTau181). Model 5: Top 14 variables identified using ShapleyVIC (four summary statistics of t-tau; minimum and mean NfL; standing and lying DBP; coffee and alcohol consumption; gender; hyperlipidemia; pTau181; and MoCA score). Model 6: Reduced set of 6 variables from ShapleyVIC-selected variables (maximum, mean, and Standard deviation of t-tau; minimum and mean NfL; and standing DBP). Model 7: Top 11 variables from XGBoost (four summary statistics of both t-tau and NfL; years of education; standing DBP; and lying DBP).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFig. 2 shows calibration results for Model 4, based on top predictors selected via RF. XGBoost and LightGBM demonstrated the best calibration, with mean absolute errors \u0026lt;0.05 and close alignment to the 45-degree reference line, indicating well-calibrated predictions. The RF model was reasonably calibrated but showed deviation in lower predicted probabilities. In contrast, KNN and NN models were poorly calibrated, underestimating risk in high-risk individuals and overestimating in low-risk cases.\u003c/p\u003e\n\u003cp\u003eIn the best-performing model\u0026mdash;XGBoost using RF-selected variables\u0026mdash;key predictors included years of education, lying and standing DBP, four summary statistics (minimum, maximum, mean, and standard deviation) for both NfL and t-tau, and p-tau181 levels. SHAP visualizations showed that elevated and variable t-tau levels strongly influenced high-risk classification. Supine and upright DBP \u0026ge;90 mmHg also contributed substantially. While NfL features were associated with cognitive decline, they were less frequently among the top three predictors. Overall, incorporating longitudinal biomarker summaries\u0026mdash;especially dynamic t-tau features\u0026mdash;improved predictive performance and enabled more parsimonious models suitable for clinical application.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this prospective study of early PD, integrating longitudinal blood biomarkers\u0026mdash;particularly t-tau and NfL\u0026mdash;significantly enhanced cognitive decline prediction using ML. The XGBoost model with RF-selected features achieved the best performance (AUC = 0.806), a substantial improvement over the baseline-only model (AUC = 0.560). Across algorithms, dynamic changes in t-tau and DBP emerged as the most consistent predictors. The consistent identification of t-tau features highlights tau pathology\u0026apos;s role in PD-related cognitive decline. Notably, t-tau variability was a strong predictor, indicating temporal fluctuations may reflect meaningful biological processes like axonal damage or inflammation.\u003c/p\u003e\n\u003cp\u003eAlthough CSF biomarkers are informative, blood-based assays offer minimal invasiveness and increasing reliability with technologies like Simoa. Our results support using serial plasma t-tau as a prognostic tool in PD, enabling longitudinal monitoring of neurodegeneration. Patients with rising or fluctuating t-tau levels could be candidates for further assessment or early intervention.\u003c/p\u003e\n\u003cp\u003eThe inclusion of NfL as a predictive biomarker also corroborates prior evidence of its value [48, 49]. NfL reflects axonal injury and has been associated with faster motor and cognitive progression in PD \u0026nbsp;[35, 50-55]. However, its predictive strength in our study was somewhat less robust than that of t-tau, suggesting that while NfL may capture broader neurodegenerative processes, tau-specific mechanisms may be more intimately tied to cognitive trajectories. These distinctions could be important in future therapeutic trials, where stratifying patients by biomarker phenotype may help tailor interventions or identify treatment responders.\u003c/p\u003e\n\u003cp\u003eBeyond molecular biomarkers, our study reinforces the prognostic importance of DBP\u0026mdash;both in lying and standing positions\u0026mdash;as a consistent clinical predictor of cognitive decline. The association of elevated DBP with two-fold increased risk has significant clinical implications [56-58]. Autonomic dysfunction is common in PD and can lead to labile blood pressure patterns, including supine hypertension and orthostatic hypotension. These fluctuations may impair cerebral autoregulation, reduce perfusion to vulnerable brain regions, and accelerate neurodegeneration. Current PD guidelines focus on motor symptoms, but our findings support routine blood pressure monitoring to identify vascular risk and guide early intervention.\u003c/p\u003e\n\u003cp\u003eImportantly, blood pressure represents a modifiable risk factor. Interventions such as midodrine, fludrocortisone, or domperidone may be used to address hypotension, while antihypertensives including ACE inhibitors or calcium channel blockers could be considered for elevated supine DBP, ideally under close monitoring to avoid worsening orthostatic symptoms. In patients with high DBP and fluctuating tau profiles, a multidisciplinary approach involving neurology, cardiology, and internal medicine may be warranted. Incorporating vascular health into PD management may not only reduce cognitive risk but also improve cardiovascular outcomes in this aging population.\u003c/p\u003e\n\u003cp\u003eAnother key finding was the protective effect of education, supporting the cognitive reserve hypothesis. Participants with \u0026ge;10 years of formal education had a 58% lower odds of cognitive decline than those with less education. While not modifiable retrospectively, educational attainment may aid in risk stratification. Individuals with limited education could benefit from closer cognitive monitoring, targeted interventions, and referral to neuropsychology. They may also be prioritized for emerging strategies to preserve cognitive function, including lifestyle modification, cognitive training, and digital therapeutics.\u003c/p\u003e\n\u003cp\u003eOur results also point to the potential value of deploying ML-based prediction tools in clinical workflows. By integrating time-varying biomarker data, clinical measures, and demographic factors, ML algorithms can generate individualized risk scores that support early identification and proactive care. For instance, a PD patient with high DBP and rising t-tau levels could be flagged for cognitive rehabilitation or clinical trial referral, while a low-risk individual might focus on motor symptom management. These tools could be embedded in electronic health records (EHRs), providing decision support to clinicians and improving consistency in care delivery. Furthermore, ML-derived insights could enhance patient-provider communication, allowing neurologists to frame prognosis in more concrete terms, support advance care planning, and tailor surveillance strategies.\u003c/p\u003e\n\u003cp\u003eIn clinical trial design, predictive models like ours can support enrichment strategies to boost statistical power and reduce sample sizes. Given the heterogeneity of PD progression, selecting high-risk individuals based on longitudinal tau and DBP dynamics may enhance trial efficiency and increase the likelihood of detecting treatment effects. This approach is particularly valuable for emerging disease-modifying therapies, such as anti-tau and anti-synuclein agents, where early intervention is key.\u003c/p\u003e\n\u003cp\u003eOur findings also align with broader evidence suggesting that mixed pathology is common in PD, particularly in those with cognitive impairment [59, 60]. While PD is defined pathologically by Lewy bodies composed of \u0026alpha;-synuclein, many patients with dementia show concurrent tau and amyloid pathology on autopsy or PET imaging [61, 62]. The observed importance of tau dynamics in our study supports this concept and may prompt reconsideration of PD as a disease spectrum rather than a single pathology [63]. This perspective opens the door to cross-disease therapeutic strategies and biomarker-guided personalization of care, particularly for those who meet overlapping criteria for PD, dementia with Lewy bodies, or AD [64, 65].\u003c/p\u003e\n\u003cp\u003eFrom a mechanistic standpoint, the interplay between vascular dysregulation and tau-driven neurodegeneration is particularly noteworthy. Chronic cerebrovascular dysfunction can disrupt glymphatic clearance of toxic proteins, promote blood-brain barrier breakdown, and create a permissive environment for tau aggregation. Conversely, tau pathology may exacerbate vascular reactivity deficits through endothelial dysfunction or oxidative stress. This bidirectional model suggests that targeting both pathways may offer synergistic benefits. For example, interventions aimed at reducing vascular stiffness or enhancing neurovascular coupling might complement disease-modifying therapies targeting tau aggregation or phosphorylation.\u003c/p\u003e\n\u003cp\u003eDespite its strengths, this study has limitations. While we used rigorous internal validation, external validation in independent cohorts is needed to confirm generalizability. The primarily Asian composition of the PALS cohort may limit applicability to other populations. Biomarkers were measured at only three time points; more frequent sampling could enhance trajectory analysis. Finally, although we focused on interpretable features and avoided overfitting, future studies should assess whether adding biomarkers of synaptic dysfunction, neuroinflammation, or \u0026alpha;-synuclein activity improves prediction. Future research should prioritize replication in multiethnic cohorts, integration of imaging biomarkers (e.g., tau PET, diffusion MRI), and development of practical risk calculators for clinical use. Cost-effectiveness studies will be essential to evaluate the feasibility of widespread biomarker monitoring. As high-sensitivity blood assays become more accessible, their implementation will be facilitated by advances in digital health, cloud analytics, and decentralized data sharing.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study shows that integrating longitudinal biomarker dynamics, particularly t-tau, with ML significantly enhances prediction of cognitive decline in early PD. These findings support the use of regular biomarker monitoring, vascular risk management, and predictive tools to enable personalized care and improve trial efficiency\u0026mdash;marking a key step toward precision neurology in PD.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u003c/strong\u003e Conceptualization, R.M., A.S.L.N., E.-K.T., L.C.S.T., and S.E.S.; Methodology, R.M., W.G., and S.E.S.; Software, R.M. and S.E.S.; Validation, R.M. and S.E.S.; Formal Analysis, R.M. and S.E.S.; Investigation, R.M. and S.E.S.; Resources, S.N., J.Y.T., A.S.L.N., X.D., X.C., D.L.H., S.N., Z.X., K.-Y.T., W.-L.A., E.-K.T., L.C.S.T., and S.E.S.; Data Curation, S.N., J.Y.T., A.S.L.N., X.D., X.C., and S.E.S.; Writing\u0026mdash;Original Draft Preparation, R.M. and S.E.S.; Writing\u0026mdash;Review and Editing, R.M., S.E.S., J.Y.T., A.S.L.N., X.D., X.C., D.L.H., S.N., Z.X., K.-Y.T., W.-L.A., E.-K.T., L.C.S.T., W.G., M.L., and S.E.S.; Visualization, R.M. and S.E.S.; Supervision, A.S.L.N., E.-K.T., L.C.S.T., W.G., M.L., and S.E.S.; Project Administration, R.M., S.Y.E.N., and J.Y.T.; Funding Acquisition, E.-K.T., L.C.S.T., and S.E.S. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This research was funded by the Singapore Ministry of Health\u0026rsquo;s National Medical Research Council (MOH-OFLCG18May-0002, MOH-CSAINV21-0005, CNIG22jul-0004).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInstitutional Review Board Statement:\u003c/strong\u003e The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of SingHealth (IRB reference number: CIRB 2019-2433) and National University of Singapore (IRB reference number: NUS-IRB-2022-899).\u003c/p\u003e\n\u003cp\u003eInformed Consent Statement: Informed consent was obtained from all subjects involved in the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement:\u003c/strong\u003e The study data will be made available upon reasonable request to the corresponding author. The data are not publicly available due to privacy and ethical concerns.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest:\u003c/strong\u003e The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eHely, M. A.; Reid, W. G. J.; Adena, M. A.; Halliday, G. M.; Morris, J. G. L., The Sydney multicenter study of Parkinson\u0026apos;s disease: the inevitability of dementia at 20 years. \u003cem\u003eMovement disorders \u003c/em\u003e\u003cstrong\u003e2008,\u003c/strong\u003e \u003cem\u003e23\u003c/em\u003e (6), 837-844.\u003c/li\u003e\n\u003cli\u003eSimuni, T.; Sethi, K., Nonmotor manifestations of Parkinson\u0026apos;s disease. \u003cem\u003eAnnals of Neurology: Official Journal of the American Neurological Association and the Child Neurology Society \u003c/em\u003e\u003cstrong\u003e2008,\u003c/strong\u003e \u003cem\u003e64\u003c/em\u003e (S2), S65-S80.\u003c/li\u003e\n\u003cli\u003eRiedel, O.; Klotsche, J.; Spottke, A.; Deuschl, G.; F\u0026ouml;rstl, H.; Henn, F.; Heuser, I.; Oertel, W.; Reichmann, H.; Riederer, P., Frequency of dementia, depression, and other neuropsychiatric symptoms in 1,449 outpatients with Parkinson\u0026rsquo;s disease. \u003cem\u003eJournal of neurology \u003c/em\u003e\u003cstrong\u003e2010,\u003c/strong\u003e \u003cem\u003e257\u003c/em\u003e, 1073-1082.\u003c/li\u003e\n\u003cli\u003eDeng, X.; Saffari, S. E.; Xiao, B.; Ng, S. Y. E.; Chia, N.; Choi, X.; Heng, D. L.; Xu, Z.; Tay, K.-Y.; Au, W.-L., Disease progression in Parkinson\u0026rsquo;s disease patients with mild cognitive impairment: 5-year longitudinal study from the early Parkinson\u0026rsquo;s disease longitudinal Singapore (PALS) cohort. \u003cem\u003eAging (Albany NY) \u003c/em\u003e\u003cstrong\u003e2024,\u003c/strong\u003e \u003cem\u003e16\u003c/em\u003e (15), 11491.\u003c/li\u003e\n\u003cli\u003eXu, F.; Huang, H.; Feng, J.; Shen, Q.; Bao, Y.; Zhang, D.; Xu, Y., Cerebrospinal fluid tau and disease progression in early Parkinson\u0026apos;s disease: an 8-year longitudinal study. \u003cem\u003eJ Neurol \u003c/em\u003e\u003cstrong\u003e2024,\u003c/strong\u003e \u003cem\u003e272\u003c/em\u003e (1), 61.\u003c/li\u003e\n\u003cli\u003eWilliams-Gray, C. H.; Mason, S. L.; Evans, J. R.; Foltynie, T.; Brayne, C.; Robbins, T. W.; Barker, R. A., The CamPaIGN study of Parkinson\u0026apos;s disease: 10-year outlook in an incident population-based cohort. \u003cem\u003eJournal of Neurology, Neurosurgery \u0026amp; Psychiatry \u003c/em\u003e\u003cstrong\u003e2013,\u003c/strong\u003e \u003cem\u003e84\u003c/em\u003e (11), 1258-1264.\u003c/li\u003e\n\u003cli\u003eGreenland, J. C.; Williams‐Gray, C. H.; Barker, R. A., The clinical heterogeneity of Parkinson\u0026apos;s disease and its therapeutic implications. \u003cem\u003eEuropean Journal of Neuroscience \u003c/em\u003e\u003cstrong\u003e2019,\u003c/strong\u003e \u003cem\u003e49\u003c/em\u003e (3), 328-338.\u003c/li\u003e\n\u003cli\u003eRoheger, M.; Kalbe, E.; Liepelt-Scarfone, I., Progression of cognitive decline in Parkinson\u0026rsquo;s disease. \u003cem\u003eJournal of Parkinson\u0026apos;s disease \u003c/em\u003e\u003cstrong\u003e2018,\u003c/strong\u003e \u003cem\u003e8\u003c/em\u003e (2), 183-193.\u003c/li\u003e\n\u003cli\u003eDeng, X.; Saffari, S. E.; Xiao, B.; Ng, S. Y. E.; Chia, N.; Choi, X.; Heng, D. L.; Ng, E.; Xu, Z.; Tay, K.-Y., Disease Progression of Data-Driven Subtypes of Parkinson\u0026rsquo;s Disease: 5-Year Longitudinal Study from the Early Parkinson\u0026rsquo;s Disease Longitudinal Singapore (PALS) Cohort. \u003cem\u003eJournal of Parkinson\u0026rsquo;s Disease \u003c/em\u003e\u003cstrong\u003e2024,\u003c/strong\u003e \u003cem\u003e14\u003c/em\u003e (5), 1051-1059.\u003c/li\u003e\n\u003cli\u003eFang, C.; Lv, L.; Mao, S.; Dong, H.; Liu, B., Cognition deficits in Parkinson\u0026rsquo;s disease: mechanisms and treatment. \u003cem\u003eParkinson\u0026rsquo;s disease \u003c/em\u003e\u003cstrong\u003e2020,\u003c/strong\u003e \u003cem\u003e2020\u003c/em\u003e (1), 2076942.\u003c/li\u003e\n\u003cli\u003ePressley, J. C.; Louis, E. D.; Tang, M. X.; Cote, L.; Cohen, P. D.; Glied, S.; Mayeux, R., The impact of comorbid disease and injuries on resource use and expenditures in parkinsonism. \u003cem\u003eNeurology \u003c/em\u003e\u003cstrong\u003e2003,\u003c/strong\u003e \u003cem\u003e60\u003c/em\u003e (1), 87-93.\u003c/li\u003e\n\u003cli\u003eShen, J.; Amari, N.; Zack, R.; Skrinak, R. T.; Unger, T. L.; Posavi, M.; Tropea, T. F.; Xie, S. X.; Van Deerlin, V. M.; Dewey Jr, R. B., Plasma MIA, CRP, and albumin predict cognitive decline in Parkinson\u0026apos;s disease. \u003cem\u003eAnnals of neurology \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e92\u003c/em\u003e (2), 255-269.\u003c/li\u003e\n\u003cli\u003eMattsson, N.; Andreasson, U.; Zetterberg, H.; Blennow, K.; Alzheimer\u0026rsquo;s Disease Neuroimaging, I., Association of plasma neurofilament light with neurodegeneration in patients with Alzheimer disease. \u003cem\u003eJAMA neurology \u003c/em\u003e\u003cstrong\u003e2017,\u003c/strong\u003e \u003cem\u003e74\u003c/em\u003e (5), 557-566.\u003c/li\u003e\n\u003cli\u003eAarsland, D.; Creese, B.; Politis, M.; Chaudhuri, K. R.; Ffytche, D. H.; Weintraub, D.; Ballard, C., Cognitive decline in Parkinson disease. \u003cem\u003eNature Reviews Neurology \u003c/em\u003e\u003cstrong\u003e2017,\u003c/strong\u003e \u003cem\u003e13\u003c/em\u003e (4), 217-231.\u003c/li\u003e\n\u003cli\u003eOizumi, H.; Hasegawa, T.; Kawahata, I.; Sekimori, T.; Totsune, T.; Sugimura, Y.; Baba, T.; Fukunaga, K.; Takeda, A., Associations among blood biomarkers, clinical subtypes, and prognosis in Parkinson\u0026rsquo;s disease. \u003cem\u003eClinical Parkinsonism \u0026amp; Related Disorders \u003c/em\u003e\u003cstrong\u003e2025,\u003c/strong\u003e \u003cem\u003e12\u003c/em\u003e, 100313.\u003c/li\u003e\n\u003cli\u003eMartin, F. L.; Williamson, S. J. M.; Paleologou, K. E.; Allsop, D.; El-Agnaf, O. M. A., \u0026alpha;-Synuclein and the pathogenesis of Parkinson\u0026apos;s disease. \u003cem\u003eProtein and Peptide Letters \u003c/em\u003e\u003cstrong\u003e2004,\u003c/strong\u003e \u003cem\u003e11\u003c/em\u003e (3), 229-237.\u003c/li\u003e\n\u003cli\u003eParnetti, L.; Gaetani, L.; Eusebi, P.; Paciotti, S.; Hansson, O.; El-Agnaf, O.; Mollenhauer, B.; Blennow, K.; Calabresi, P., CSF and blood biomarkers for Parkinson\u0026apos;s disease. \u003cem\u003eThe Lancet Neurology \u003c/em\u003e\u003cstrong\u003e2019,\u003c/strong\u003e \u003cem\u003e18\u003c/em\u003e (6), 573-586.\u003c/li\u003e\n\u003cli\u003eYoussef, P.; Hughes, L.; Kim, W. S.; Halliday, G. M.; Lewis, S. J. G.; Cooper, A.; Dzamko, N., Evaluation of plasma levels of NFL, GFAP, UCHL1 and tau as Parkinson\u0026apos;s disease biomarkers using multiplexed single molecule counting. \u003cem\u003eSci Rep \u003c/em\u003e\u003cstrong\u003e2023,\u003c/strong\u003e \u003cem\u003e13\u003c/em\u003e (1), 5217.\u003c/li\u003e\n\u003cli\u003eKhalil, M.; Teunissen, C. E.; Otto, M.; Piehl, F.; Sormani, M. P.; Gattringer, T.; Barro, C.; Kappos, L.; Comabella, M.; Fazekas, F., Neurofilaments as biomarkers in neurological disorders. \u003cem\u003eNature Reviews Neurology \u003c/em\u003e\u003cstrong\u003e2018,\u003c/strong\u003e \u003cem\u003e14\u003c/em\u003e (10), 577-589.\u003c/li\u003e\n\u003cli\u003eB\u0026auml;ckstr\u0026ouml;m, D. C.; Domell\u0026ouml;f, M. E.; Linder, J.; Olsson, B.; \u0026Ouml;hrfelt, A.; Trupp, M.; Zetterberg, H.; Blennow, K.; Forsgren, L., Cerebrospinal fluid patterns and the risk of future dementia in early, incident Parkinson disease. \u003cem\u003eJAMA neurology \u003c/em\u003e\u003cstrong\u003e2015,\u003c/strong\u003e \u003cem\u003e72\u003c/em\u003e (10), 1175-1182.\u003c/li\u003e\n\u003cli\u003eLerche, S.; Wurster, I.; R\u0026ouml;ben, B.; Zimmermann, M.; Machetanz, G.; Wiethoff, S.; Dehnert, M.; Rietschel, L.; Riebenbauer, B.; Deuschle, C., CSF NFL in a longitudinally assessed PD cohort: age effects and cognitive trajectories. \u003cem\u003eMovement Disorders \u003c/em\u003e\u003cstrong\u003e2020,\u003c/strong\u003e \u003cem\u003e35\u003c/em\u003e (7), 1138-1144.\u003c/li\u003e\n\u003cli\u003eBacioglu, M.; Maia, L. F.; Preische, O.; Schelle, J.; Apel, A.; Kaeser, S. A.; Schweighauser, M.; Eninger, T.; Lambert, M.; Pilotto, A., Neurofilament light chain in blood and CSF as marker of disease progression in mouse models and in neurodegenerative diseases. \u003cem\u003eNeuron \u003c/em\u003e\u003cstrong\u003e2016,\u003c/strong\u003e \u003cem\u003e91\u003c/em\u003e (1), 56-66.\u003c/li\u003e\n\u003cli\u003eLin, C.-H.; Li, C.-H.; Yang, K.-C.; Lin, F.-J.; Wu, C.-C.; Chieh, J.-J.; Chiu, M.-J., Blood NfL: a biomarker for disease severity and progression in Parkinson disease. \u003cem\u003eNeurology \u003c/em\u003e\u003cstrong\u003e2019,\u003c/strong\u003e \u003cem\u003e93\u003c/em\u003e (11), e1104-e1111.\u003c/li\u003e\n\u003cli\u003eMollenhauer, B.; Bibl, M.; Wiltfang, J.; Steinacker, P.; Ciesielczyk, B.; Neubert, K.; Trenkwalder, C.; Otto, M., Total tau protein, phosphorylated tau (181p) protein, \u0026beta;-amyloid1\u0026ndash;42, and \u0026beta;-amyloid1\u0026ndash;40 in cerebrospinal fluid of patients with dementia with Lewy bodies. De Gruyter: 2006.\u003c/li\u003e\n\u003cli\u003eSiderowf, A.; Xie, S. X.; Hurtig, H.; Weintraub, D.; Duda, J.; Chen-Plotkin, A.; Shaw, L. M.; Van Deerlin, V.; Trojanowski, J. Q.; Clark, C., CSF amyloid \u0026beta; 1-42 predicts cognitive decline in Parkinson disease. \u003cem\u003eNeurology \u003c/em\u003e\u003cstrong\u003e2010,\u003c/strong\u003e \u003cem\u003e75\u003c/em\u003e (12), 1055-1061.\u003c/li\u003e\n\u003cli\u003eChen, N.-C.; Chen, H.-L.; Li, S.-H.; Chang, Y.-H.; Chen, M.-H.; Tsai, N.-W.; Yu, C.-C.; Yang, S.-Y.; Lu, C.-H.; Lin, W.-C., Plasma levels of \u0026alpha;-synuclein, A\u0026beta;-40 and T-tau as biomarkers to predict cognitive impairment in Parkinson\u0026rsquo;s disease. \u003cem\u003eFrontiers in aging neuroscience \u003c/em\u003e\u003cstrong\u003e2020,\u003c/strong\u003e \u003cem\u003e12\u003c/em\u003e, 112.\u003c/li\u003e\n\u003cli\u003eAamodt, W. W.; Waligorska, T.; Shen, J.; Tropea, T. F.; Siderowf, A.; Weintraub, D.; Grossman, M.; Irwin, D.; Wolk, D. A.; Xie, S. X.; Trojanowski, J. Q.; Shaw, L. M.; Chen-Plotkin, A. S., Neurofilament Light Chain as a Biomarker for Cognitive Decline in Parkinson Disease. \u003cem\u003eMov Disord \u003c/em\u003e\u003cstrong\u003e2021,\u003c/strong\u003e \u003cem\u003e36\u003c/em\u003e (12), 2945-2950.\u003c/li\u003e\n\u003cli\u003eMa, L. Z.; Zhang, C.; Wang, H.; Ma, Y. H.; Shen, X. N.; Wang, J.; Tan, L.; Dong, Q.; Yu, J. T., Serum Neurofilament Dynamics Predicts Cognitive Progression in de novo Parkinson\u0026apos;s Disease. \u003cem\u003eJ Parkinsons Dis \u003c/em\u003e\u003cstrong\u003e2021,\u003c/strong\u003e \u003cem\u003e11\u003c/em\u003e (3), 1117-1127.\u003c/li\u003e\n\u003cli\u003eChen‐Plotkin, A. S.; Hu, W. T.; Siderowf, A.; Weintraub, D.; Goldmann Gross, R.; Hurtig, H. I.; Xie, S. X.; Arnold, S. E.; Grossman, M.; Clark, C. M., Plasma epidermal growth factor levels predict cognitive decline in Parkinson disease. \u003cem\u003eAnnals of neurology \u003c/em\u003e\u003cstrong\u003e2011,\u003c/strong\u003e \u003cem\u003e69\u003c/em\u003e (4), 655-663.\u003c/li\u003e\n\u003cli\u003eChiu, M.-J.; Yang, S.-Y.; Chen, T.-F.; Lin, C.-H.; Yang, F.-C.; Chen, W.-P.; Zetterberg, H.; Blennow, K., Synergistic association between plasma A\u0026beta;1\u0026ndash;42 and p-tau in Alzheimer\u0026rsquo;s disease but not in Parkinson\u0026rsquo;s disease or frontotemporal dementia. \u003cem\u003eACS Chemical Neuroscience \u003c/em\u003e\u003cstrong\u003e2021,\u003c/strong\u003e \u003cem\u003e12\u003c/em\u003e (8), 1376-1383.\u003c/li\u003e\n\u003cli\u003eDing, J.; Zhang, J.; Wang, X.; Zhang, L.; Jiang, S.; Yuan, Y.; Li, J.; Zhu, L.; Zhang, K., Relationship between the plasma levels of neurodegenerative proteins and motor subtypes of Parkinson\u0026rsquo;s disease. \u003cem\u003eJournal of neural transmission \u003c/em\u003e\u003cstrong\u003e2017,\u003c/strong\u003e \u003cem\u003e124\u003c/em\u003e, 353-360.\u003c/li\u003e\n\u003cli\u003eWong, Y. Y.; Wu, C.-Y.; Yu, D.; Kim, E.; Wong, M.; Elez, R.; Zebarth, J.; Ouk, M.; Tan, J.; Liao, J., Biofluid markers of blood-brain barrier disruption and neurodegeneration in Lewy body spectrum diseases: A systematic review and meta-analysis. \u003cem\u003eParkinsonism \u0026amp; Related Disorders \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e101\u003c/em\u003e, 119-128.\u003c/li\u003e\n\u003cli\u003eBatzu, L.; Rota, S.; Hye, A.; Heslegrave, A.; Trivedi, D.; Gibson, L. L.; Farrell, C.; Zinzalias, P.; Rizos, A.; Zetterberg, H., Plasma p-tau181, neurofilament light chain and association with cognition in Parkinson\u0026rsquo;s disease. \u003cem\u003enpj Parkinson\u0026apos;s Disease \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e8\u003c/em\u003e (1), 154.\u003c/li\u003e\n\u003cli\u003eSanchez, E.; Wilkinson, T.; Coughlan, G.; Mirza, S.; Baril, A. A.; Ramirez, J.; Binns, M. A.; Black, S. E.; Borrie, M.; Dilliott, A. A.; Dixon, R. A.; Dowlatshahi, D.; Farhan, S.; Finger, E.; Fischer, C. E.; Frank, A.; Freedman, M.; Goncalves, R. A.; Grimes, D. A.; Hassan, A.; Hegele, R. A.; Kumar, S.; Lang, A. E.; Marras, C.; McLaughlin, P. M.; Orange, J. B.; Pasternak, S. H.; Pollock, B. G.; Rajji, T. K.; Roberts, A. C.; Robinson, J. F.; Rogaeva, E.; Sahlas, D. J.; Saposnik, G.; Strong, M. J.; Swartz, R. H.; Tang-Wai, D. F.; Tartaglia, M. C.; Troyer, A. K.; Kvartsberg, H.; Zetterberg, H.; Munoz, D. P.; Masellis, M., Association of plasma biomarkers with cognition, cognitive decline, and daily function across and within neurodegenerative diseases: Results from the Ontario Neurodegenerative Disease Research Initiative. \u003cem\u003eAlzheimers Dement \u003c/em\u003e\u003cstrong\u003e2024,\u003c/strong\u003e \u003cem\u003e20\u003c/em\u003e (3), 1753-1770.\u003c/li\u003e\n\u003cli\u003ePedersen, C. C.; Ushakova, A.; Alves, G.; Tysnes, O.-B.; Blennow, K.; Zetterberg, H.; Maple-Gr\u0026oslash;dem, J.; Lange, J., Serum neurofilament light at diagnosis: a prognostic indicator for accelerated disease progression in Parkinson\u0026rsquo;s Disease. \u003cem\u003enpj Parkinson\u0026apos;s Disease \u003c/em\u003e\u003cstrong\u003e2024,\u003c/strong\u003e \u003cem\u003e10\u003c/em\u003e (1), 162.\u003c/li\u003e\n\u003cli\u003eVijiaratnam, N.; Lawton, M.; Heslegrave, A. J.; Guo, T.; Tan, M.; Jabbari, E.; Real, R.; Woodside, J.; Grosset, K.; Chelban, V.; Athauda, D.; Girges, C.; Barker, R. A.; Hardy, J.; Wood, N.; Houlden, H.; Williams, N.; Ben-Shlomo, Y.; Zetterberg, H.; Grosset, D. G.; Foltynie, T.; Morris, H. R., Combining biomarkers for prognostic modelling of Parkinson\u0026apos;s disease. \u003cem\u003eJ Neurol Neurosurg Psychiatry \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e93\u003c/em\u003e (7), 707-15.\u003c/li\u003e\n\u003cli\u003eMattsson, N.; Cullen, N. C.; Andreasson, U.; Zetterberg, H.; Blennow, K., Association between longitudinal plasma neurofilament light and neurodegeneration in patients with Alzheimer disease. \u003cem\u003eJAMA neurology \u003c/em\u003e\u003cstrong\u003e2019,\u003c/strong\u003e \u003cem\u003e76\u003c/em\u003e (7), 791-799.\u003c/li\u003e\n\u003cli\u003eYoon, S.; Kim, M.; Lee, W.-W., Long short-term memory-based deep learning models for screening Parkinson\u0026rsquo;s disease using sequential diagnostic codes. \u003cem\u003eJournal of Clinical Neurology (Seoul, Korea) \u003c/em\u003e\u003cstrong\u003e2023,\u003c/strong\u003e \u003cem\u003e19\u003c/em\u003e (3), 270.\u003c/li\u003e\n\u003cli\u003eZhang, J.; Zhang, X.; Sh, Y.; Liu, B.; Hu, Z., Diagnostic AI Modeling and Pseudo Time Series Profiling of AD and PD Based on Individualized Serum Proteome Data. \u003cem\u003eFrontiers in Bioinformatics \u003c/em\u003e\u003cstrong\u003e2021,\u003c/strong\u003e \u003cem\u003e1\u003c/em\u003e, 764497.\u003c/li\u003e\n\u003cli\u003eLam, S.; Arif, M.; Song, X.; Uhl\u0026eacute;n, M.; Mardinoglu, A., Machine learning analysis reveals biomarkers for the detection of neurological diseases. \u003cem\u003eFrontiers in Molecular Neuroscience \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e15\u003c/em\u003e, 889728.\u003c/li\u003e\n\u003cli\u003eV\u0026aacute;squez-Correa, J. C.; Arias-Vergara, T.; Orozco-Arroyave, J. R.; Eskofier, B.; Klucken, J.; N\u0026ouml;th, E., Multimodal assessment of Parkinson\u0026apos;s disease: a deep learning approach. \u003cem\u003eIEEE journal of biomedical and health informatics \u003c/em\u003e\u003cstrong\u003e2018,\u003c/strong\u003e \u003cem\u003e23\u003c/em\u003e (4), 1618-1630.\u003c/li\u003e\n\u003cli\u003eJunaid, M.; Ali, S.; Eid, F.; El-Sappagh, S.; Abuhmed, T., Explainable machine learning models based on multimodal time-series data for the early detection of Parkinson\u0026rsquo;s disease. \u003cem\u003eComputer Methods and Programs in Biomedicine \u003c/em\u003e\u003cstrong\u003e2023,\u003c/strong\u003e \u003cem\u003e234\u003c/em\u003e, 107495.\u003c/li\u003e\n\u003cli\u003eMohammadi, R.; Ng, S. Y. E.; Tan, J. Y.; Ng, A. S. L.; Deng, X.; Choi, X.; Heng, D. L.; Neo, S.; Xu, Z.; Tay, K. Y.; Au, W. L.; Tan, E. K.; Tan, L. C. S.; Steyerberg, E. W.; Greene, W.; Saffari, S. E., Machine Learning for Early Detection of Cognitive Decline in Parkinson\u0026apos;s Disease Using Multimodal Biomarker and Clinical Data. \u003cem\u003eBiomedicines \u003c/em\u003e\u003cstrong\u003e2024,\u003c/strong\u003e \u003cem\u003e12\u003c/em\u003e (12).\u003c/li\u003e\n\u003cli\u003eNg, S. Y.-E.; Chia, N. S.-Y.; Abbas, M. M.; Saffari, E. S.; Choi, X.; Heng, D. L.; Xu, Z.; Tay, K.-Y.; Au, W.-L.; Tan, E.-K., Physical activity improves anxiety and apathy in early Parkinson\u0026apos;s disease: a longitudinal follow-up study. \u003cem\u003eFrontiers in Neurology \u003c/em\u003e\u003cstrong\u003e2021,\u003c/strong\u003e \u003cem\u003e11\u003c/em\u003e, 625897.\u003c/li\u003e\n\u003cli\u003eYong, A. C. W.; Tan, Y. J.; Zhao, Y.; Lu, Z.; Ng, E. Y. L.; Ng, S. Y. E.; Chia, N. S. Y.; Choi, X.; Heng, D.; Neo, S., SNCA Rep1 microsatellite length influences non-motor symptoms in early Parkinson\u0026rsquo;s disease. \u003cem\u003eAging (Albany NY) \u003c/em\u003e\u003cstrong\u003e2020,\u003c/strong\u003e \u003cem\u003e12\u003c/em\u003e (20), 20880.\u003c/li\u003e\n\u003cli\u003eStekhoven, D. J., Using the missForest package. \u003cem\u003eR package \u003c/em\u003e\u003cstrong\u003e2011\u003c/strong\u003e, 1-11.\u003c/li\u003e\n\u003cli\u003eYouden, W. J., Index for rating diagnostic tests. \u003cem\u003eCancer \u003c/em\u003e\u003cstrong\u003e1950,\u003c/strong\u003e \u003cem\u003e3\u003c/em\u003e (1), 32-35.\u003c/li\u003e\n\u003cli\u003eWang, X.; Yang, X.; He, W.; Song, X.; Zhang, G.; Niu, P.; Chen, T., The association of serum neurofilament light chains with early symptoms related to Parkinson\u0026apos;s disease: A cross-sectional study. \u003cem\u003eJournal of Affective Disorders \u003c/em\u003e\u003cstrong\u003e2023,\u003c/strong\u003e \u003cem\u003e343\u003c/em\u003e, 144-152.\u003c/li\u003e\n\u003cli\u003eWelton, T.; Tan, Y. J.; Saffari, S. E.; Ng, S. Y.; Chia, N. S.; Yong, A. C.; Choi, X.; Heng, D. L.; Shih, Y.-C.; Hartono, S., Plasma neurofilament light concentration is associated with diffusion-tensor MRI-based measures of neurodegeneration in early Parkinson\u0026rsquo;s disease. \u003cem\u003eJournal of Parkinson\u0026apos;s Disease \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e12\u003c/em\u003e (7), 2135-2146.\u003c/li\u003e\n\u003cli\u003eNg, A. S. L.; Tan, Y. J.; Yong, A. C. W.; Saffari, S. E.; Lu, Z.; Ng, E. Y.; Ng, S. Y. E.; Chia, N. S. Y.; Choi, X.; Heng, D., Utility of plasma Neurofilament light as a diagnostic and prognostic biomarker of the postural instability gait disorder motor subtype in early Parkinson\u0026rsquo;s disease. \u003cem\u003eMolecular neurodegeneration \u003c/em\u003e\u003cstrong\u003e2020,\u003c/strong\u003e \u003cem\u003e15\u003c/em\u003e, 1-8.\u003c/li\u003e\n\u003cli\u003eNg, A. S. L.; Tan, Y. J.; Yong, A. C. W.; Saffari, S. E.; Lu, Z.; Ng, E. Y.; Ng, S. Y. E.; Chia, N. S. Y.; Choi, X.; Heng, D.; Neo, S.; Xu, Z.; Keong, N. C. H.; Tay, K. Y.; Au, W. L.; Tan, L. C. S.; Tan, E. K., Utility of plasma Neurofilament light as a diagnostic and prognostic biomarker of the postural instability gait disorder motor subtype in early Parkinson\u0026apos;s disease. \u003cem\u003eMol Neurodegener \u003c/em\u003e\u003cstrong\u003e2020,\u003c/strong\u003e \u003cem\u003e15\u003c/em\u003e (1), 33.\u003c/li\u003e\n\u003cli\u003eYgland R\u0026ouml;dstr\u0026ouml;m, E.; Mattsson-Carlgren, N.; Janelidze, S.; Hansson, O.; Puschmann, A., Serum neurofilament light chain as a marker of progression in Parkinson\u0026rsquo;s disease: Long-term observation and implications of clinical subtypes. \u003cem\u003eJournal of Parkinson\u0026apos;s Disease \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e12\u003c/em\u003e (2), 571-584.\u003c/li\u003e\n\u003cli\u003eHalloway, S.; Desai, P.; Beck, T.; Aggarwal, N.; Agarwal, P.; Evans, D.; Rajan, K. B., Association of Neurofilament Light With the Development and Severity of Parkinson Disease. \u003cem\u003eNeurology \u003c/em\u003e\u003cstrong\u003e2022,\u003c/strong\u003e \u003cem\u003e98\u003c/em\u003e (22), e2185-e2193.\u003c/li\u003e\n\u003cli\u003eOu, R.; Liu, K.; Lin, J.; Yang, T.; Xiao, Y.; Wei, Q.; Hou, Y.; Li, C.; Zhang, L.; Jiang, Z.; Zhao, B.; Chen, X.; Song, W.; Wu, Y.; Shang, H., Relationship between plasma NFL and disease progression in Parkinson\u0026apos;s disease: a prospective cohort study. \u003cem\u003eJ Neurol \u003c/em\u003e\u003cstrong\u003e2024,\u003c/strong\u003e \u003cem\u003e271\u003c/em\u003e (4), 1837-1843.\u003c/li\u003e\n\u003cli\u003eMollenhauer, B.; Dakna, M.; Kruse, N.; Galasko, D.; Foroud, T.; Zetterberg, H.; Schade, S.; Gera, R. G.; Wang, W.; Gao, F., Validation of serum neurofilament light chain as a biomarker of Parkinson\u0026apos;s disease progression. \u003cem\u003eMovement Disorders \u003c/em\u003e\u003cstrong\u003e2020,\u003c/strong\u003e \u003cem\u003e35\u003c/em\u003e (11), 1999-2008.\u003c/li\u003e\n\u003cli\u003eKwon, K.-Y.; Pyo, S. J.; Lee, H. M.; Seo, W.-K.; Koh, S.-B., Cognition and visit-to-visit variability of blood pressure and heart rate in de novo patients with Parkinson\u0026rsquo;s disease. \u003cem\u003eJournal of Movement Disorders \u003c/em\u003e\u003cstrong\u003e2016,\u003c/strong\u003e \u003cem\u003e9\u003c/em\u003e (3), 144.\u003c/li\u003e\n\u003cli\u003eDoiron, M.; Langlois, M.; Dupr\u0026eacute;, N.; Simard, M., The influence of vascular risk factors on cognitive function in early Parkinson\u0026apos;s disease. \u003cem\u003eInternational Journal of Geriatric Psychiatry \u003c/em\u003e\u003cstrong\u003e2018,\u003c/strong\u003e \u003cem\u003e33\u003c/em\u003e (2), 288-297.\u003c/li\u003e\n\u003cli\u003eXiao, Y.; Yang, T.; Zhang, L.; Wei, Q.; Ou, R.; Hou, Y.; Liu, K.; Lin, J.; Jiang, Q.; Shang, H., Association between the blood pressure variability and cognitive decline in Parkinson\u0026apos;s disease. \u003cem\u003eBrain Behav \u003c/em\u003e\u003cstrong\u003e2023,\u003c/strong\u003e \u003cem\u003e13\u003c/em\u003e (12), e3319.\u003c/li\u003e\n\u003cli\u003eIrwin, D. J.; Grossman, M.; Weintraub, D.; Hurtig, H. I.; Duda, J. E.; Xie, S. X.; Lee, E. B.; Van Deerlin, V. M.; Lopez, O. L.; Kofler, J. K., Neuropathological and genetic correlates of survival and dementia onset in synucleinopathies: a retrospective analysis. \u003cem\u003eThe Lancet Neurology \u003c/em\u003e\u003cstrong\u003e2017,\u003c/strong\u003e \u003cem\u003e16\u003c/em\u003e (1), 55-65.\u003c/li\u003e\n\u003cli\u003eRobinson, J. L.; Lee, E. B.; Xie, S. X.; Rennert, L.; Suh, E.; Bredenberg, C.; Caswell, C.; Van Deerlin, V. M.; Yan, N.; Yousef, A., Neurodegenerative disease concomitant proteinopathies are prevalent, age-related and APOE4-associated. \u003cem\u003eBrain \u003c/em\u003e\u003cstrong\u003e2018,\u003c/strong\u003e \u003cem\u003e141\u003c/em\u003e (7), 2181-2193.\u003c/li\u003e\n\u003cli\u003eGomperts, S. N.; Locascio, J. J.; Makaretz, S. J.; Schultz, A.; Caso, C.; Vasdev, N.; Sperling, R.; Growdon, J. H.; Dickerson, B. C.; Johnson, K., Tau positron emission tomographic imaging in the Lewy body diseases. \u003cem\u003eJAMA neurology \u003c/em\u003e\u003cstrong\u003e2016,\u003c/strong\u003e \u003cem\u003e73\u003c/em\u003e (11), 1334-1341.\u003c/li\u003e\n\u003cli\u003eCompta, Y.; Parkkinen, L.; O\u0026apos;Sullivan, S. S.; Vandrovcova, J.; Holton, J. L.; Collins, C.; Lashley, T.; Kallis, C.; Williams, D. R.; de Silva, R., Lewy-and Alzheimer-type pathologies in Parkinson\u0026apos;s disease dementia: which is more important? \u003cem\u003eBrain \u003c/em\u003e\u003cstrong\u003e2011,\u003c/strong\u003e \u003cem\u003e134\u003c/em\u003e (5), 1493-1505.\u003c/li\u003e\n\u003cli\u003eJellinger, K. A., Dementia with Lewy bodies and Parkinson\u0026rsquo;s disease-dementia: current concepts and controversies. \u003cem\u003eJournal of neural transmission \u003c/em\u003e\u003cstrong\u003e2018,\u003c/strong\u003e \u003cem\u003e125\u003c/em\u003e, 615-650.\u003c/li\u003e\n\u003cli\u003eMcKeith, I. G.; Boeve, B. F.; Dickson, D. W.; Halliday, G.; Taylor, J.-P.; Weintraub, D.; Aarsland, D.; Galvin, J.; Attems, J.; Ballard, C. G., Diagnosis and management of dementia with Lewy bodies: Fourth consensus report of the DLB Consortium. \u003cem\u003eNeurology \u003c/em\u003e\u003cstrong\u003e2017,\u003c/strong\u003e \u003cem\u003e89\u003c/em\u003e (1), 88-100.\u003c/li\u003e\n\u003cli\u003eEspay, A. J.; Schwarzschild, M. A.; Tanner, C. M.; Fernandez, H. H.; Simon, D. K.; Leverenz, J. B.; Merola, A.; Chen‐Plotkin, A.; Brundin, P.; Kauffman, M. A., Biomarker‐driven phenotyping in Parkinson\u0026apos;s disease: a translational missing link in disease‐modifying clinical trials. \u003cem\u003eMovement Disorders \u003c/em\u003e\u003cstrong\u003e2017,\u003c/strong\u003e\u003cem\u003e32\u003c/em\u003e (3), 319-324.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"npj-parkinsons-disease","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjparkd","sideBox":"Learn more about [npj Parkinson's Disease](http://www.nature.com/npjparkd/)","snPcode":"41531","submissionUrl":"https://submission.springernature.com/new-submission/41531/3","title":"npj Parkinson's Disease","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7107548/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7107548/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eCognitive decline is a major non-motor complication in early Parkinson\u0026rsquo;s disease (PD), but predicting its progression remains challenging. Using data from 193 participants in the Early Parkinson\u0026rsquo;s Disease Longitudinal Singapore (PALS) cohort, we evaluated whether longitudinal blood biomarkers\u0026mdash;neurofilament light chain (NfL) and total tau (t-tau)\u0026mdash;could improve prediction of cognitive decline, defined as a one-point annual or sustained two-year drop in Montreal Cognitive Assessment scores. We applied three variable selection methods and five machine learning models across seven feature sets. Overall, 23% of participants experienced cognitive decline over five years. The XGBoost model trained on Random Forest\u0026ndash;selected variables achieved the highest performance (AUC\u0026thinsp;=\u0026thinsp;0.806), a substantial improvement over the baseline-only model (AUC\u0026thinsp;=\u0026thinsp;0.560). Key predictors included diastolic blood pressure and summaries of t-tau and NfL. Time-varying biomarkers improved predictions over baseline data alone, supporting their integration with machine learning for early cognitive risk assessment in PD.\u003c/p\u003e","manuscriptTitle":"Machine Learning Integration of Serial Blood Biomarkers Enhances Cognitive Decline Prediction in Early Parkinson's Disease","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-28 07:24:57","doi":"10.21203/rs.3.rs-7107548/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-15T05:15:10+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-15T01:24:00+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-01T16:09:47+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"254742194171042703361266192790510141860","date":"2025-07-22T17:08:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"253632966785411443835169597861280199519","date":"2025-07-22T11:45:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"306267924181147656607872614601142989709","date":"2025-07-22T11:37:45+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-22T09:31:08+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-15T20:43:57+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-15T17:07:47+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Parkinson's Disease","date":"2025-07-12T10:29:22+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-parkinsons-disease","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjparkd","sideBox":"Learn more about [npj Parkinson's Disease](http://www.nature.com/npjparkd/)","snPcode":"41531","submissionUrl":"https://submission.springernature.com/new-submission/41531/3","title":"npj Parkinson's Disease","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"b5bd89a2-1796-4b06-abd4-98bf9344f990","owner":[],"postedDate":"July 28th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":51949244,"name":"Health sciences/Biomarkers"},{"id":51949245,"name":"Health sciences/Neurology"},{"id":51949246,"name":"Biological sciences/Neuroscience"}],"tags":[],"updatedAt":"2026-03-02T16:03:07+00:00","versionOfRecord":{"articleIdentity":"rs-7107548","link":"https://doi.org/10.1038/s41531-026-01298-8","journal":{"identity":"npj-parkinsons-disease","isVorOnly":false,"title":"npj Parkinson's Disease"},"publishedOn":"2026-02-25 15:58:13","publishedOnDateReadable":"February 25th, 2026"},"versionCreatedAt":"2025-07-28 07:24:57","video":"","vorDoi":"10.1038/s41531-026-01298-8","vorDoiUrl":"https://doi.org/10.1038/s41531-026-01298-8","workflowStages":[]},"version":"v1","identity":"rs-7107548","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7107548","identity":"rs-7107548","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}