Figure 2.

Absolute risk distribution and performance of the BCRAT and IBIS models in the clinical-based cohort. (A) Distribution of models’ predictions as a function of cumulative percentage of women. Rug plot on the y-axis. (B) Calibration according to the models’ predictions groups. P values were computed using a goodness-of-fit test statistic compared with the critical value from the χ² distribution. (C) Discrimination power of the models according to sensitivity and specificity. C-index was calculated using the inverse probability of censoring weighting estimation of cumulative time-dependent ROC curve. BCRAT, Breast Cancer Risk Assessment Tool; E/O, expected-to-observed cases; IBIS, International Breast Cancer Intervention Study; ROC, receiver operating characteristic.