Table 4.
Model | Group | Clinical diagnosis (n) | Classified as non-TB (n) | Classified as TB (n) | Cut-off value | Sensitivity (95% CI) |
Specificity (95% CI) |
PPV (95% CI) |
NPV (95% CI) |
Accuracy (95% CI) |
Training (n=276) |
Non-TB | 189 | 189 | 0 | 0.4 | 100% (96% to 100%) |
100% (98% to 100%) |
100% (98% to 100%) |
100% (96% to 100%) |
100% (99% to 100%) |
TB | 87 | 0 | 87 | |||||||
Test (n=92) |
Non-TB | 63 | 52 | 11 | 0.4 | 76% (56% to 90%) |
83% (71% to 91%) |
88% (77% to 95%) |
67% (48% to 82%) |
80% (71% to 88%) |
TB | 29 | 7 | 22 | |||||||
Prediction (n=102) |
Non-TB | 70 | 59 | 11 | 0.4 | 75% (57% to 89%) |
84% (74% to 92%) |
88% (78% to 95%) |
69% (51% to 83%) |
81% (72% to 88%) |
TB | 32 | 8 | 24 |
NPV, negative predictive value; PPV, positive predictive value.