Fig 1.
Pathways of humanin.
Fig 2.
Pathways of MOTS-c.
Fig 3.
Pathways of p66shc.
Table 1.
Depression severity level with corresponding PHQ-9 scores
Fig 4.
Correlation matrix of all twelve features.
Fig 5.
Correlation matrix of features after the dropping HT-MedUse and DM-MedUse.
Table 2.
Class distribution for 5-class classification.
Table 3.
Class distribution for 3-class classification.
Table 4.
Class distribution for binary classes.
Fig 6.
Flowchart for predicting depression severity.
Table 5.
ML algorithms used in this study.
Fig 7.
Model 1: All Variables for binary classification.
Fig 8.
Model 1: All Variables for 3-class classification.
Fig 9.
Model 1: All Variables for 5-class classification.
Fig 10.
Model 2: Biomarkers + ACE + Age + Gender.
Fig 11.
Model 3: Biomarkers only.
Fig 12.
SHAP feature importance for Model 1 (binary classification).
Fig 13.
The SHAP Summary plot of Model 1’s Feature effects.
Fig 14.
The SHAP Waterfall Plot for Model 1’s Prediction for Instance 6.
Fig 15.
The LIME explanation for Instance 20.
Fig 16.
The LIME explanation for Instance 70.
Table 6.
Comparison of the performance evaluation of Random Forest results for binary classification on both balanced and unbalanced data set.
Table 7.
Comparison of the performance evaluation of Random Forest results for 3-class classification on both balanced and unbalanced data set.
Table 8.
Comparison of the performance evaluation of Random Forest results for 5-class classification on both balanced and unbalanced data set.