Table 2.
Accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) values for each machine learning model.
| Accuracy (%) | Sensitivity (%) | Specificity (%) | AUC | |
|---|---|---|---|---|
| Decision tree | 54.7 ± 1.5 | 53.3 ± 2.0 | 54.9 ± 1.7 | 0.562 ± 0.015 |
| Majority model | 61.9 ± 0.8 | 55.4 ± 1.1 | 69.2 ± 1.3 | 0.568 ± 0.009 |
| Random forest | 57.2 ± 0.8 | 54.4 ± 1.2 | 60.4 ± 1.1 | 0.615 ± 0.007 |
| SVM (linear) | 61.4 ± 0.5 | 57.1 ± 0.6 | 66.7 ± 0.8 | 0.622 ± 0.002 |
| SVM (non-linear) | 61.9 ± 0.4 | 52.3 ± 1.5 | 71.6 ± 1.1 | 0.623 ± 0.005 |
| Confidence model | 61.5 ± 0.9 | 49.1 ± 1.4 | 67.1 ± 1.0 | 0.633 ± 0.008 |
| Logistic regression | 59.1 ± 0.5 | 55.5 ± 0.6 | 62.6 ± 0.8 | 0.635 ± 0.001 |
| k-Nearest neighbor | 61.8 ± 0.6 | 46.6 ± 1.0 | 72.1 ± 0.8 | 0.641 ± 0.004 |
| Neural network | 62.0 ± 0.9 | 53.3 ± 1.3 | 71.2 ± 1.9 | 0.646 ± 0.005 |
All data are mean ± SD; SVM, Support Vector Machine.