Table 2.
Performance of machine learning models.
| Model | AUC | ACC (%) | Youden index (%) | Sensitivity (%) | Specificity (%) | F1 score | PPV (%) | NPV (%) |
|---|---|---|---|---|---|---|---|---|
| XGBoost | 0.90 | 81 | 70 | 89 | 81 | 0.26 | 15 | 99 |
| CatBoost | 0.88 | 80 | 65 | 86 | 80 | 0.24 | 14 | 99 |
| LightGBM | 0.85 | 84 | 57 | 73 | 85 | 0.25 | 15 | 99 |
| MLP | 0.80 | 80 | 47 | 67 | 80 | 0.19 | 11 | 98 |
| SVM | 0.78 | 74 | 47 | 73 | 74 | 0.17 | 10 | 99 |
| LR | 0.74 | 67 | 40 | 73 | 67 | 0.14 | 8 | 98 |
| Random forest | 0.74 | 71 | 40 | 69 | 71 | 0.15 | 8 | 98 |
| Gradient boosting | 0.71 | 37 | 34 | 100 | 34 | 0.10 | 5 | 100 |
| KNN | 0.66 | 74 | 29 | 55 | 75 | 0.13 | 8 | 98 |
| AdaBoost | 0.61 | 85 | 27 | 40 | 87 | 0.16 | 10 | 97 |
| Naive Bayes | 0.59 | 38 | 21 | 86 | 36 | 0.09 | 5 | 98 |
XGBOOST, eXtremely Gradient Boosting; CatBoost, Categorical Boosting; LightGBM, Light Gradient Boosting; MLP, Multi-Layer Perceptron; SVM, Support Vector Machine; LR, Logistic Regression. KNN, K-Nearest Neighbor; AdaBoost, Adaptive boosting; ACC, accuracy, PPV, positive predictive value; NPV, negative predictive value.