Fig. 3From: Integrating multiple molecular sources into a clinical risk prediction signature by extracting complementary informationPrediction error curves for the first AML data application example. Bootstrap.632+ prediction error curves estimates for sequential complementary strategy, i.e. boosting including SNP information for adjustment (dashed red curve) and for the reference approach, i.e. boosting without including SNP information (solid blue curve). The Kaplan-Meier benchmark is indicated by the dashed-dotted gray curve and the Cox model is given by dotted black curveBack to article page