Fig. 3From: Statistical models versus machine learning for competing risks: development and validation of prognostic modelsMiscalibration of cause-specific Cox model, Fine-Gray model, PLANNCR original, PLANNCR extended (tuned with Brier score at 5 years), and RSFCR at 2, 5, and 10 years for the event of interest: disease progression based on 100 validation datasets. Miscalibration was calculated as the mean squared error (MSE) between the observed and the predicted cumulative incidence event probabilities (for 4 groups)Back to article page