scen_param.RdThe scenarios were derived so that the prognostic part of the model on the control arm has an R^2 of 0.32 for continuous data and an AUC of 0.66 for binary data. The coefficient b0 was calculated in each case so that the overall test for a treatment effect has a power of 50 (defined by the expressions for prognostic and predictive part) replicated across continuous and binary endpoints. Within each scenario there are 5 sub-scenarios corresponding to different selections of b1. The third case in each of the sub-scenarios correspond to the case where the interaction test (under the true model) has 80 b1 > 0 is considered). The other 4 cases correspond to 0, 0.5, 1.5, 2 times the b1 value that provides 80 test. The column b1_rel gives the size of b1 relative to the b1 value that provides 80 scenarios.
scen_paramSee more details on the choice of parameters considered in https://pubmed.ncbi.nlm.nih.gov/36437036/
See more details on the the data cases in https://pubmed.ncbi.nlm.nih.gov/36437036/