Table 2:
Discrimination indicators of the predictive models in the training and external validation sets.
| Models | AUC (95% CI) | Youden index | ACC (%) | Sensitivity (%) | Specificity (%) | F1 score | PPV (%) | NPV (%) |
|---|---|---|---|---|---|---|---|---|
| Training set | ||||||||
| LR | 0.788 (0.767–0.808) | 0.45 | 72.42 | 64.64 | 80.19 | 0.70 | 76.54 | 69.40 |
| XGBoost | 0.941 (0.931–0.952) | 0.75 | 87.49 | 93.50 | 81.47 | 0.88 | 83.46 | 92.62 |
| RF | 0.894 (0.880–0.908) | 0.61 | 80.51 | 85.62 | 75.40 | 0.81 | 77.68 | 83.99 |
| External validation set | ||||||||
| LR | 0.691 (0.613–0.769) | 0.23 | 59.66 | 40.62 | 82.50 | 0.52 | 73.58 | 53.66 |
| XGBoost | 0.724 (0.648–0.800) | 0.19 | 56.82 | 32.29 | 86.25 | 0.45 | 73.81 | 51.49 |
| RF | 0.677 (0.599–0.756) | 0.23 | 59.66 | 42.71 | 80.00 | 0.54 | 71.93 | 53.78 |