Table 3 Performance of models built based on different ML algorithms.
Data sets | Models | Accuracy (95%CI) | Sensitivity (95%CI) | Specificity (95%CI) | F1 score (95%CI) |
|---|---|---|---|---|---|
Training data | LR | 0.821(0.806–0.837) | 0.873(0.843–0.904) | 0.798(0.763–0.833) | 0.753(0.743–0.764) |
LightGBM | 0.800(0.759–0.840) | 0.759(0.649–0.870) | 0.818(0.756–0.880) | 0.700(0.634–0.767) | |
GNB | 0.810(0.791–0.828) | 0.851(0.820–0.882) | 0.791(0.751–0.830) | 0.737(0.724–0.749) | |
XGBoost | 0.964(0.948–0.979) | 0.970(0.962–0.977) | 0.961(0.937–0.984) | 0.944(0.921–0.966) | |
AdaBoost | 0.899(0.883–0.915) | 0.926(0.901–0.951) | 0.887(0.859–0.915) | 0.851(0.833–0.870) | |
SVM | 0.865(0.860–0.871) | 0.880(0.855–0.904) | 0.859(0.840–0.878) | 0.803(0.799–0.806) | |
Validation data | LR | 0.801(0.770–0.832) | 0.848(0.797–0.899) | 0.779(0.716–0.842) | 0.728(0.703–0.753) |
LightGBM | 0.765(0.747–0.784) | 0.702(0.585–0.819) | 0.794(0.744–0.844) | 0.647(0.602–0.692) | |
GNB | 0.795(0.772–0.818) | 0.817(0.749–0.885) | 0.785(0.757–0.813) | 0.712(0.678–0.746) | |
XGBoost | 0.876(0.854–0.897) | 0.822(0.717–0.926) | 0.899(0.848–0.950) | 0.802(0.762–0.843) | |
AdaBoost | 0.864(0.831–0.897) | 0.867(0.816–0.917) | 0.862(0.817–0.908) | 0.8(0.756–0.844) | |
SVM | 0.811(0.784–0.837) | 0.816(0.699–0.932) | 0.808(0.766–0.850) | 0.725(0.668–0.781) |
