Fig. 6From: Weighted Cox regression for the prediction of heterogeneous patient subgroupsEstimated effects in the application. Different types of Cox models including the top-1000-variance genes as covariates. Mean estimated regression coefficients of selected genes (averaged across all training sets regardless of whether a gene is selected in individual splits or not). For each subgroup genes with a mean inclusion frequency larger or equal than 0.5 in any model type are selectedBack to article page