ANC Workshop - 21st January 2025

Title: Targeted Learning for Real-World Evidence

Abstract: 

The National Institute for Heath and Care Excellence (NICE) defines Real-World Data as “data relating to patient health or experience or care delivery collected outside the context of a highly controlled clinical trial.”, i.e., observational data whereby interventions or exposures are not determined by a study protocol but by patients and healthcare professional. In this talk, I will discuss the Targeted Learning (TL) Roadmap for Real-World Evidence generation. TL is a methodology unifying machine learning, mathematical statistics and causal inference, with theoretical guarantees for performance. I will discuss how Targeted Maximum Likelihood Estimation reduces/annihilate model-misspecification bias, demonstrate this through simulations, and present two of our recent works where TL estimators have been developed and/or applied for applications in large-scale genomic and biomarker discovery studies using the UK Biobank.