Table 4:
The Brier score, sensitivity, and specificity of all five models at the smallest percentage of diseased cases in the population. Note that the numbers in the parentheses are percentages of diseased cases included in the training sets.
| Hospital Discharge dataset (0.5%) | Myocardial Infarction dataset (2%) | |||||
|---|---|---|---|---|---|---|
| Brier Score | Sensitivity | Specificity | Brier Score | Sensitivity | Specificity | |
| SVM | 0.3684 | 0.6765 | 0.5370 | 0.4018 | 0.6536 | 0.4450 |
| RSVM | 0.2898 | 0.6634 | 0.5830 | 0.3460 | 0.6307 | 0.5067 |
| LR | 0.0812 | 0.6386 | 0.5560 | 0.1800 | 0.2320 | 0.8544 |
| Morik’s algorithm | 0.1160 | 0.6766 | 0.5673 | 0.3472 | 0.5948 | 0.5453 |
| PE-LR | 0.0803 | 0.7464 | 0.5737 | 0.1765 | 0.7674 | 0.4916 |