Table 2.
Machine learning models performance
| Models | AUC (95% CI) | AUPRC (95% CI) | AUPRC baseline | Gmean |
|---|---|---|---|---|
| LightGBM | 0.88 (0.83–0.92) | 0.42 (0.36–0.48) | 0.12 | 0.48 |
| XGBoost | 0.91 (0.87–0.95) | 0.49 (0.43–0.55) | 0.12 | 0.76 |
| Random Forest | 0.89 (0.85–0.93) | 0.54 (0.46–0.61) | 0.12 | 0.39 |
| Glmnet | 0.86 (0.78–0.94) | 0.47 (0.43–0.52) | 0.12 | 0.54 |
AUC Area Under the Curve, AUPRC Area Under the Precision-Recall Curve. Gmean means the sqrt (sensitivity * specificity). 95% CI shows the uncertainty for AUC and AUPRC metrics. AUPRC baseline means the value that a random classifier would achieve (random guessing)