Table 2. Logistic Regression and Decision Tree Models performance by cohort.
| Model | Dataset | Sensitivity (95% CI) | Specificity (95% CI) | Accuracy (95% CI) |
|---|---|---|---|---|
| Logistic Regression | Development Cohort | 100% (74.65 to 100) | 93% (64.2 to 99.6) | 97% (82.2 to 99.9) |
| Validation Cohort | 90% (54.1 to 99.5) | 62% (32.3 to 84.9) | 74% (51.3 to 88.9) | |
| All | 96% (77.7 to 99.8) | 78% (57.3 to 90.6) | 87% (73.6 to 93.9) | |
| Decision Tree | Development Cohort | 93% (66 to 99.6) | 93% (64.2 to 99.6) | 93% (75.8 to 98.8) |
| Validation Cohort | 90% (54.1 to 99.5) | 62% (32.3 to 84.9) | 74% (51.3 to 87.4) | |
| All | 92% (72.5 to 98.6) | 78% (57.3 to 90.6) | 85% (71.4 to 92.7) |
The lower and upper limits of the 95% confidence interval (CI) for a proportion were calculated according to Newcombe, using the Wilson procedure with a correction for continuity [35].