Table 2.
Diagnostic performance of the ANN
| Sensitivity (TP/TP + FN) | 95% CI | Specificity (TN/TN + FP) | 95% CI | +LR | −LR | |
|---|---|---|---|---|---|---|
| Training set (n = 313) | ||||||
| C1 | 100% (207/207) | 98.2–100% | 9.4% (10/106) | 5.2–16.5% | 1.1 | 0 |
| C2 | 95.2% (197/207) | 91.3–97.7% | 42.5% (45/106) | 33.5–52.0% | 1.7 | 0.1 |
| Test set (n = 157) | ||||||
| C1 | 100% (88/88) | 95.9–100% | 14.5% (10/69) | 7.2–25% | 1.2 | 0 |
| C2 | 95.5% (84/88) | 88.8–98.7% | 36.2% (25/69) | 25.0–48.7% | 1.5 | 0.1 |
ANN artificial neural network, TP true positive, TN true negative, FP false positive, FN false negative, +/−; LR likelihood ratio, C1, 100% sensitivity cutoff; C2, > 95% sensitivity cutoff