Skip to main content

Table 1 One random split into training and test data for the Mainz cohort study

From: Survival models with preclustered gene groups as covariates

Method Covariates p LR p PI p IBS λ sel.cov
L 1 genes 0.0190 0.0017 0.1042 11.72 19
L 1 GO 0.0176 0.0018 0.1103 10.75 16
L 1 clustered 0.0092 0.0002 0.0830 28.53 5
L 2 genes 0.0098 0.0003 0.0877 5112.08 17834
L 2 GO 0.0541 0.0097 0.1022 11749.16 6530
L 2 clustered 0.0690 0.0006 0.0896 96499.04 31229
  1. Results for the two prediction methods using (i) genes, (ii) GO groups, and (iii) preclustered GO groups. For ridge regression, nearly all covariates are kept in the model since parameter estimates are unlikely to get shrunken exactly to 0.LR = ^ logranktest, PI = ^ prognosticindex,IBS = ^ IntegratedBrierScore,sel.cov = ^ numberofselectedcovariates.