Table 3.
Scores for each model in the testing set.
| Models | AUC (95% CI) | Specificity | Sensitivity | Precision | Accuracy |
|---|---|---|---|---|---|
| LR | 0.857 (0.814–0.900) | 0.912 | 0.620 | 0.761 | 0.821 |
| SVM | 0.865 (0.823–0.907) | 0.912 | 0.602 | 0.756 | 0.816 |
| RFC | 0.862 (0.820–0.904) | 0.883 | 0.657 | 0.717 | 0.813 |
| XGBoost | 0.858 (0.815–0.901) | 0.895 | 0.630 | 0.731 | 0.813 |
| DNN | 0.867 (0.827–0.908) | 0.891 | 0.556 | 0.811 | 0.821 |
| LR* | 0.866 (0.825–0.907) | 0.921 | 0.593 | 0.780 | 0.821 |
| RFC* | 0.874 (0.835–0.912) | 0.950 | 0.500 | 0.818 | 0.810 |
AUC, the area under the curve; LR, logistic regression; SVM, support vector machine; RFC, random forest classifier; XGBoost, extreme gradient boosting; DNN, fully-connected deep neural network.
*
indicates model developed with 21 variables.