Table 3 Predictive performance of 7 ML models in training and validation sets.
Sets | Classifiers | Accuracy (95%CI) | Sensitivity (95%CI) | Specificity (95%CI) | F1 score (95%CI) |
|---|---|---|---|---|---|
Training set | XGBoost | 0.947(0.914–0.979) | 0.942(0.904–0.979) | 0.954(0.925–0.983) | 0.953(0.925–0.982) |
LR | 0.740(0.729–0.751) | 0.657(0.637–0.676) | 0.857(0.848–0.867) | 0.747(0.733–0.760) | |
LightGBM | 0.746(0.710–0.783) | 0.737(0.683–0.790) | 0.760(0.708–0.812) | 0.771(0.733–0.808) | |
GNB | 0.713(0.709–0.718) | 0.643(0.609–0.677) | 0.813(0.774–0.852) | 0.723(0.710–0.736) | |
MLP | 0.694(0.666–0.723) | 0.675(0.613–0.737) | 0.722(0.641–0.803) | 0.719(0.689–0.748) | |
SVM | 0.806(0.788–0.824) | 0.767(0.734–0.801) | 0.861(0.852–0.871) | 0.822(0.800–0.843) | |
AdaBoost | 0.825(0.814–0.837) | 0.845(0.826–0.865) | 0.797(0.771–0.824) | 0.850(0.840–0.860) | |
Validation set | XGBoost | 0.830(0.798–0.863) | 0.848(0.807–0.888) | 0.806(0.750–0.862) | 0.853(0.824–0.883) |
LR | 0.729(0.694–0.764) | 0.644(0.602–0.685) | 0.850(0.794–0.906) | 0.734(0.697–0.772) | |
LightGBM | 0.720(0.679–0.762) | 0.761(0.688–0.834) | 0.663(0.592–0.734) | 0.758(0.716–0.800) | |
GNB | 0.691(0.663–0.718) | 0.619(0.566–0.673) | 0.790(0.759–0.822) | 0.698(0.661–0.735) | |
MLP | 0.666(0.613–0.719) | 0.654(0.579–0.729) | 0.684(0.561–0.807) | 0.694(0.646–0.743) | |
SVM | 0.753(0.727–0.779) | 0.720(0.669–0.770) | 0.800(0.736–0.864) | 0.772(0.743–0.800) | |
AdaBoost | 0.759(0.722–0.797) | 0.806(0.762–0.850) | 0.694(0.641–0.747) | 0.796(0.761–0.830) |
