Genetic epidemiology and clinical studies

The first aim is to derive practical recommendations for identifying interactions between predictive biomarkers and treatments in different study designs via simulations based on distinct regression models. The first step is to design simulation scenarios based on effects and effect combinations described in literature. Next, transformations are developed to define equivalent effects in different regression models resulting in subsequent simulation runs. The second aim is to give recommendations how to analyse subgroup designs defined by multiple biomarkers. The first step here is again to design simulation scenarios by literature search. Practical recommendations arise from simulations based on these scenarios, and unbiased estimators for treatment and biomarker effects will be developed.