Model performance at several classification thresholds of predicted probability
| Threshold | Sensitivity | Specificity | PPV | NPV | LR+ | LR− |
| 0.121* | 65 (59–70) | 74 (72–75) | 24 (21–27) | 94 (93–95) | 2.4 (1.6–2.9) | 0.5 (0.4–0.6) |
| 0.075 | 81 (76–85) | 44 (42–46) | 15 (14–17) | 95 (93–96) | 1.4 (1.0–1.6) | 0.4 (0.4–0.5) |
| 0.067 | 85 (80–88) | 38 (36–40) | 15 (13–17) | 95 (94–96) | 1.4 (1.2–1.6) | 0.4 (0.2–0.5) |
| 0.047 | 90 (86–93) | 22 (20–24) | 13 (12–14) | 95 (92–96) | 1.2 (0.9–1.2) | 0.4 (0.3–0.6) |
| 0.034 | 95 (92–97) | 11 (10–13) | 12 (11–13) | 95 (92–97) | 1.1 (1.0–1.1) | 0.4 (0.3–0.6) |
Data are presented for the single data set used for model selection. Numbers in brackets represent the 95% CIs.
*The ‘optimal’ threshold according to Youden’s J statistic.
LR+, positive likelihood ratio; LR−, negative likelihood ratio; NPV, negative predictive value; PPV, positive predictive value;