Table 2. Discrimination and calibration performance of machine learning models in differentiating lung cancer from controls.
| Model | AUC (95% CI) | Accuracy | F1 score | NPV | PPV | Sensitivity | Specificity | Brier score | Threshold |
|---|---|---|---|---|---|---|---|---|---|
| LASSO | 0.926 (0.834–1.000) | 0.957 | 0.963 | 0.966 | 0.951 | 0.975 | 0.933 | 0.069 | 0.646 |
| LightGBM | 0.956 (0.882–1.000) | 0.914 | 0.923 | 0.875 | 0.947 | 0.900 | 0.933 | 0.074 | 0.577 |
| SVM | 0.947 (0.878–1.000) | 0.943 | 0.950 | 0.933 | 0.950 | 0.950 | 0.933 | 0.075 | 0.582 |
| XGBoost | 0.951 (0.895–1.000) | 0.929 | 0.937 | 0.903 | 0.949 | 0.975 | 0.933 | 0.074 | 0.493 |
Notes.
- AUC
- area under the receiver operating characteristic curve
- Accuracy
- proportion of correctly classified cases
- Sensitivity
- true positive rate
- Specificity
- true negative rate
- PPV
- positive predictive value
- NPV
- negative predictive value
- F1 score
- harmonic mean of sensitivity and PPV
- Brier score
- mean squared error between predicted probabilities and observed outcomes (range: 0–1, with lower values indicating better calibration)