Fig. 3: Feature importance based on SHAP values for the stacking ensemble machine learning model trained on 49 quantified metabolites and 25 predicted clinical parameters.

Features are ranked from most to least influential based on their average absolute SHAP values. Each point represents a SHAP value for an individual observation, with color indicating the original feature value (red = high, blue = low). A positive SHAP value indicates that the feature contributes to a higher predicted age, while a negative SHAP value indicates a contribution to a lower predicted age. This plot illustrates both the magnitude and direction of each feature’s impact on model predictions.