Model specific tests | Model A | Model B | Model C |
---|---|---|---|
Mean pred. probability of RRP cases/non-cases | 9.8/9.1 | 29.6/20.2 | 19.9/18.3 |
C-statistics (95% CI) | 0.59 (0.50, 0.67) | 0.65 (0.58, 0.73) | 0.57 (0.49, 0.64) |
Hosmer-Lemeshow p-value (χ2/d.f.) | 0.059 (10.6/5) | 0.005 (22.0/8) | 0.047 (9.64/4) |
Comparison between models | Model A vs. Model B | Model C vs. Model B | Model A vs. Model C |
IDI‡ (p-value) | 8.7 (<0.001) | 7.3 (0.005) | 0.9 (0.25) |
‡IDI compares the difference in mean predicted probability between cases and non-cases for two models. A positive number indicates better performance of the comparator model (mentioned last in the column heading).