Table 3 Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and balanced accuracy for prespecified high-risk thresholds for a logistic regression model containing established risk factors only (Old model), a logistic regression model containing both established and additional risk factors (New model), and ensemble model in the test dataset.
Threshold | Sensitivity | Specificity | PPV | NPV | Balanced accuracy | |
|---|---|---|---|---|---|---|
0.1/10% | Old model | 0.67 (0.63, 0.71) | 0.74 (0.73, 0.75) | 0.19 (0.18, 0.21) | 0.96 (0.95, 0.97) | 0.704 |
New model | 0.67 (0.63, 0.70) | 0.74 (0.73, 0.75) | 0.19 (0.18, 0.21) | 0.96 (0.95, 0.96) | 0.704 | |
Ensemble model* | 0.65 (0.61, 0.69) | 0.75 (0.73, 0.76) | 0.19 (0.18, 0.21) | 0.96 (0.95, 0.96) | 0.697 | |
0.2/20% | Old model | 0.29 (0.26, 0.33) | 0.93 (0.92, 0.93) | 0.27 (0.24, 0.31) | 0.93 (0.93, 0.94) | 0.610 |
New model | 0.31 (0.27, 0.35) | 0.92 (0.92, 0.93) | 0.28 (0.24, 0.31) | 0.93 (0.93, 0.94) | 0.617 | |
Ensemble model* | 0.23 (0.20, 0.26) | 0.96 (0.95, 0.96) | 0.33 (0.29, 0.38) | 0.93 (0.92, 0.94) | 0.593 | |
