PURPOSE: Individual and population health can be improved by better targeting of treatments. This personalisation requires an understanding of how treatment effectiveness differs across patient subgroups. However, in RCTs the interpretation of subgroup analysis is challenging due to the risk of spurious “false positive” results arising from multiple testing, compounded by lack of power. The analysis of large observational data sets, and the application of biomics are becoming important but still pose significant analytic challenges. We will explore some advanced methods to address these issues. DESCRIPTION: The workshop comprises three talks on the following interconnected themes: 1) an introduction to subgroup analysis, and the opportunities raised by developments in biomics and big data for decision-modelling 2) the use of fully Bayesian methods (including prior elicitation, shrinkage and mixture models) for subgroup analysis of RCTs 3) an econometric approach that draws on concepts of causal inference to estimate person-level treatment effects from large observational datasets. Each talk will discuss the pros and cons of the various methods and will include: examples of how the method can be applied, and will focus on each method's ability to detect ‘true’ heterogeneity’, and to ensure that results are reproducible. Workshop participants will be actively engaged through Q&A and polling before and after each short talk. The workshop will identify lessons for the design, analysis and interpretation of future studies, and future research and training needs.