Table 3 Performance of CAD4TB (v6), Lunit (v4.7.2), and qXR (v2) at Selected Thresholds.
Thresholds | CAD4TB (v6) | Lunit (v4.7.2) | qXR (v2) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Threshold | Accuracy | Sensitivity (95% CI) | Specificity (95% CI) | Threshold | Accuracy | Sensitivity (95% CI) | Specificity (95% CI) | Threshold | Accuracy | Sensitivity (95% CI) | Specificity (95% CI) | |
ROC01# | 63 | 0.85 | 0.91 (0.84–0.96) | 0.84 (0.82–0.86) | 0.92 | 0.89 | 0.87 (0.79–0.93) | 0.89 (0.87–0.91) | 0.67 | 0.86 | 0.88 (0.8–0.93) | 0.89 (0.87–0.91) |
Sensitivity ≥95% | 57 | 0.81 | 0.95 (0.9–0.98) | 0.8 (0.77–0.82) | 0.55 | 0.77 | 0.95 (0.9–0.98) | 0.76 (0.73–0.78) | 0.49 | 0.83 | 0.95 (0.90–0.98) | 0.82 (0.79–0.84) |
Reduce Xpert tests by 1/2 | 47 | 0.61 | 0.97 (0.92–0.99) | 0.57 (0.54–0.6) | 0.11 | 0.6 | 0.99 (0.95–1) | 0.56 (0.53–0.59) | 0.18 | 0.57 | 0.97 (0.92–0.99) | 0.53 (0.5–0.56) |
Reduce Xpert tests by 2/3 | 53 | 0.75 | 0.96 (0.91–0.99) | 0.73 (0.7–0.76) | 0.39 | 0.74 | 0.95 (0.9–0.98) | 0.72 (0.69–0.74) | 0.34 | 0.75 | 0.96 (0.91–0.99) | 0.73 (0.7–0.76) |
Reduce Xpert tests by 3/4 | 59 | 0.82 | 0.94 (0.87–0.97) | 0.82 (0.79–0.84) | 0.79 | 0.82 | 0.93 (0.86–0.97) | 0.81 (0.79–0.84) | 0.5 | 0.83 | 0.93 (0.86–0.97) | 0.82 (0.8–0.84) |
Max Accuracy | 88 | 0.92 | 0.47 (0.37–0.57) | 0.96 (0.95–0.97) | 0.98 | 0.94 | 0.58 (0.48–0.67) | 0.97 (0.96–0.98) | 0.84 | 0.94 | 0.71 (0.61–0.79) | 0.96 (0.94–0.97) |
