Fig. 1.
Study design, methodology, and analyses. A The general structure of the Colombian Survey on Health, Well-Being, and Aging (SHWA). The top left panel depicts the Colombian regions (Central, Pacific, Amazonian, Atlantic, and Bogota regions) where data were collected to complete 23,694 individuals. The middle panel in A shows the different types of factors assessed in the study. The predictors included in the theory-driven approach (DG, SDH, and CMF) are detained in the red boxes. Moreover, additional health variables factors were assessed in the data-driven approach (blue boxes). The top right panel in A shows the two outcomes assessed in this study (cognitive and functionality levels). B A simplified model of the theory-driven approach using the multigroup structural equation modelling procedure. C The machine learning procedures used in the data-driven analyses. These analyses involve the implementation of a XGBoost procedure and progressive feature elimination procedure. DG, socio-demographic conditions; SDH, social determinants of health; CMF, cardiometabolic factors; MH, mental health factors; PF, physical functioning; LF, lifestyle factors