Figure 2.

Performance of the model. (A,B) Classification plots of the models with and without CRP for development (A) and validation (B). (C) Calibration plots of the models with and without CRP using 10-fold cross-validation with the total training dataset. (D) Calibration plot of the models with and without CRP on the external validation dataset. Each datapoint represents 10% of the data presented as mean ± 95% confidence intervals. The diagonal white line represents the ideal calibration line with an intercept of 0 and a regression coefficient of 1. (E, F) Based on the risk scores provided by the algorithm without CRP, the patients were separated into four risk groups, and the proportion of patients in each group are displayed.