Table 5.
Performance of each algorithm in validation cohort using incident events included in hospital or mortality data
| Incident events | Mean (95% CI) | |
|---|---|---|
| Women | Men | |
| Upper gastrointestinal bleed: | ||
| ROC statistic* | 0.766 (0.758 to 0.775) | 0.747 (0.738 to 0.756) |
| R2 (%)† | 40.7 (38.9 to 42.6) | 36.9 (35.1 to 38.7) |
| D statistic‡ | 1.70 (1.63 to 1.76) | 1.57 (1.51 to 1.63) |
| Intracranial bleed: | ||
| ROC statistic* | 0.847 (0.838 to 0.856) | 0.812 (0.80 to 0.824) |
| R2 (%)† | 58.0 (56.0 to 60.0) | 53.3 (51.1 to 55.4) |
| D statistic‡ | 2.40 (2.30 to 2.50) | 2.19 (2.09 to 2.28) |
Discrimination is ability of risk prediction model to differentiate between patients who do and do not experience an event during the study.
*Receiver operating characteristic (ROC) curve statistic; higher values indicate better discrimination.
†Measures explained variation, with higher values indicating more variation explained.
‡Measure of discrimination specific to censored survival data. As with ROC, higher values indicate better discrimination.