Table 5.
Test performance of the algorithm.
| Aimed sensitivity level | Sensitivity (Actual) (%) | Specificity (%) | Positive predictive value (%) | Negative predictive value (%) | Balanced accuracy | F1-score |
|---|---|---|---|---|---|---|
| Sensitivity of 1 | 100 | 40.2 | 48.3 | 100.0 | 0.701 | 0.651 |
| Sensitivity of 0.975 | 97.5 | 49.6 | 51.9 | 97.2 | 0.735 | 0.677 |
| Sensitivity of 0.95 | 94.9 | 53.0 | 53.0 | 94.9 | 0.739 | 0.680 |
| Sensitivity of 0.90 | 88.8 | 67.4 | 60.3 | 91.5 | 0.781 | 0.719 |
| Sensitivity of 0.85 | 84.3 | 73.1 | 63.6 | 89.3 | 0.787 | 0.725 |
| Sensitivity of 0.80 | 79.2 | 79.6 | 68.4 | 87.3 | 0.794 | 0.734 |
| Sensitivity of 0.75 | 71.6 | 84.1 | 71.6 | 84.1 | 0.779 | 0.716 |
| Best balanced accuracy | 77.7 | 79.9 | 68.3 | 86.5 | 0.788 | 0.727 |
AUC, Area Under the Receiving Operating Curve.