Causal Roadmap for generation of reliable Real-World Evidence with application to genomics and biomarker discovery
We will discuss the importance of following the Causal Roadmap applied to Real-World Data (RWD) for generation of robust Real-World Evidence (RWE). We will demonstrate the integration of machine learning methods with semi-parametric efficient estimation methodologies for double-robust inference of target quantities of interest, which has recently become of great interest for Health Technology Assessment by regulatory bodies. Examples of double robust strategies will be presented with concrete applications to biomarker discovery and genomics. We will explain the importance of reducing model misspecification bias, especially for scenarios with large sample size and small effect size such as genomics, in prioritising interventions or candidates for experimental validation.
