Skip to main content
Fig. 8 | BMC Cancer

Fig. 8

From: Spatially varying effects of predictors for the survival prediction of nonmetastatic colorectal Cancer

Fig. 8

Model calibration performance based on prediction error curves. The nonparametric Kaplan-Meier curve was used as a reference. Cox and AFT represent the statistical model, and RSF represents the machine learning model. The prediction error was estimated at 0, 12, 36, 60, 72, 96, and 108 months after diagnosis. As the prediction error curves of Cox and AFT were nearly overlapping, we plotted their curves separately. a) Prediction error curves for Cox and RSF. b) Prediction error curves for AFT and RSF

Back to article page