Fig. 1

Overview of Experimental design. Figure 1. A Data Collection. (A1) Dataset selection: MHATS and NHANES. (A2) Collection of cognitive performance measures and demographic/medical history information (A3) Acquisition and preprocessing of PA data. B Construction of Multimodal Machine Learning Models. (B1) Extraction of 777 PA features and dimensionality reduction to 20 based on RFE. (B2) Fusion of PA features and demographic features based on CCA. (B3) Modeling of fused features using 5 machine learning algorithms and hyperparameter optimization based on GridSearchCV. C Model Analysis. (C1) Visual analysis of PA data. (C2) Feature contribution analysis based on SHAP and correlation analysis based on RCS (C3) Deployment of models to the cloud. D Development of Mobile System (D1) Mobile phone interface: used for filling in basic information and connecting devices. (D2) Smartwatch interface: used for monitoring control and result push notifications