Fig. 4
Panels (A) and (B) illustrate the feature selection process for all-cause mortality based on the Borutaalgorithm and the evolution of Z-scores during the screening process, respectively. Panels (C) and (D) depict thefeature selection process for cardiovascular mortality and the corresponding Z-score evolution during screening.In Panels (A) and (C), the horizontal axis represents the variable names, while the vertical axis indicates theZ-scores for each variable. In Panels (B) and (D), the horizontal axis shows the number of iterations, and the verticalaxis reflects the changes in Z-scores during the screening process. Blue boxes and lines correspond to theminimum, mean, and maximum Z-scores of shadow features. Green boxes and lines represent variablesconfirmed by the model, while red boxes and lines indicate variables that were rejected.Panels (E) and (F) display the importance of various variables in predicting all-cause and cardiovascular mortality.
