Table 4 Performance of each model on the testing cohort.
AUC (95% CI) | ACC (%) | Sensitivity (%) | Specificity (%) | NPV (%) | PPV (%) | |
|---|---|---|---|---|---|---|
Bagging-ensemble learning model | ||||||
RF | 0.747 (0.606–0.864) | 76.7 | 31.6 | 97.6 | 75.5 | 85.7 |
Boosting-ensemble learning model | ||||||
LightGBM | 0.781 (0.654–0.889) | 66.7 | 36.8 | 80.5 | 73.3 | 46.7 |
XGBoost | 0.787 (0.661–0.888) | 70 | 31.6 | 87.8 | 73.5 | 54.5 |
AdaBoost | 0.786 (0.643–0.899) | 68.3 | 63.2 | 70.7 | 80.6 | 50 |
CatBoost | 0.804 (0.676–0.902) | 75 | 21.1 | 100 | 73.2 | 100 |
Voting-ensemble learning model | ||||||
Soft voting | 0.875 (0.765–0.953) | 80 | 47.4 | 95.1 | 79.6 | 81.8 |
