Table 3 Performance criteria of ML methods for polypharmacy prediction
Model | Set | Sensitivity (Recall)% | Specificity% | Accuracy% | Precision% | F1-score% |
|---|---|---|---|---|---|---|
DT | Train | 63.76 (0.82) | 88.67 (0.37) | 79.46 (0.38) | 63.76 (0.82) | 63.76 (0.82) |
Test | 63.69 (1.92) | 88.70 (0.86) | 79.50 (0.89) | 63.69 (1.92) | 63.69 (1.92) | |
XGBoost | Train | 66.00 (0.97) | 92.26 (0.45) | 82.55 (0.38) | 66.00 (0.97) | 66.00 (0.97) |
Test | 62.31 (1.97) | 90.20 (1.00) | 79.94 (0.80) | 62.31 (1.97) | 62.31 (1.97) | |
SVM | Train | 63.80 (0.84) | 88.97 (0.39) | 79.66 (0.38) | 63.80 (0.84) | 63.80 (0.84) |
Test | 63.46 (1.95) | 88.81 (0.86) | 79.49 (0.90) | 63.46 (1.95) | 63.46 (1.95) | |
ANN’s | Train | 27.63 (6.70) | 99.01 (1.06) | 72.61 (1.82) | 27.63 (6.70) | 27.63 (6.70) |
Test | 26.52 (7.53) | 98.82 (1.27) | 72.23 (2.18) | 26.52 (7.53) | 26.52 (7.53) | |
RF | Train | 69.85 (1.37) | 92.87 (0.42) | 84.34 (0.46) | 69.85 (1.37) | 69.85 (1.37) |
Test | 63.92 (2.27) | 89.92 (1.14) | 79.99 (0.88) | 63.92 (2.27) | 63.92 (2.27) |