Figure 1
From: Derivation and assessment of risk prediction models using case-cohort data
Mean error in 10-year predicted absolute risk. Figure shows the difference between risk predictions obtained using a prediction model derived on the case-cohort set, and those obtained using the model derived on the full cohort (case-cohort minus full cohort estimates). Solid lines show mean error, averaged over all individuals in the case-cohort set, and averaged over simulations, for three case-cohort weighting methods. Dashed lines show upper and lower fifth percentiles of the distribution of average differences across 1000 simulations.