TABLE 6.
Testing data results.
| Model | TP | TN | FP | FN | Sensitivity | Specificity | Accuracy | Balanced accuracy | AUROC |
|---|---|---|---|---|---|---|---|---|---|
| 4-layer DNN (Proposed) | 3 | 30 | 8 | 1 | 0.75 | 0.7894 | 0.7857 | 0.7697 | 0.90 |
| LR | 2 | 21 | 17 | 2 | 0.5 | 0.4473 | 0.4523 | 0.4736 | 0.47 |
| SVM | 2 | 24 | 14 | 2 | 0.5 | 0.6315 | 0.6190 | 0.5657 | 0.55 |
| KNN | 2 | 37 | 7 | 2 | 0.5 | 0.8157 | 0.7857 | 0.6578 | 0.65 |
| RF | 2 | 27 | 11 | 2 | 0.5 | 0.7105 | 0.6904 | 0.6052 | 0.55 |
| AdaBoost | 1 | 33 | 5 | 3 | 0.25 | 0.8684 | 0.8095 | 0.5592 | 0.58 |
| XGBoost | 2 | 33 | 5 | 2 | 0.5 | 0.8684 | 0.8139 | 0.6842 | 0.67 |
TP, true positives; TN, true negatives; FP, false positives; FN, false negatives; AUROC, area under receiver operating characteristics; DNN, deep neural network; DT, decision tree; SVM, support vector machine; KNN, K-Nearest Neighbor algorithm; RF, random forest; AdaBoost, adaptive boosting; XGBoost, extreme gradient boosting.