Table 4. Metrics used to compare the algorithms.
| Metric | Formula | Range | Target value |
|---|---|---|---|
| Sensitivity | TP/(TP + FN) | [0, 1] | ∼1 |
| Specificity | TN/(FP + TN) | [0, 1] | ∼1 |
| LR+ | Sensitivity/(1-specificity) | [0, +∞) | >10 |
| LR− | (1-sensitivity)/specificity | [0, +∞) | <0.1 |
| DOR | (TP/FN)/(FP/TN) | [0, +∞) | >1 |
| F1-score | TP/(TP + 1/2 (FP + FN)) | [0, 1] | ∼1 |
| AUROC | Area under the ROC curve | [0, 1] | ∼1 |
Note:
TP, True positives; TN, True negatives; FP, False positives; FN, False negatives.