Data Diversity X Stasitical Methods Tutorial - What Works for Whom: Assessing Heterogeneous Treatment Effects (1:25:55)

About the video:
Averages can be misleading: a treatment can look effective overall while doing little, or even causing harm, for large parts of the population it's meant to help. Heterogeneous treatment effects capture that variation — how a treatment's benefit differs from person to person, rather than just its average across a trial population. Understanding who benefits (and who doesn't) is central to delivering more equitable care.

On September 10th 2026, Dr Carly Brantner (Duke University) led a 90-minute interactive tutorial on detecting, estimating, and interpreting that variation — live-coding in R on a simulated dataset based on real data comparing treatments for depression, moving from traditional approaches (interaction terms, stratified analysis) through to more advanced techniques such as causal forests. It saw firsthand what each method reveals, and what it trades off, when it comes to identifying who a treatment actually works for.

This session was a partnership between DSxHE's Statistical Methods and Data Diversity Themes, in collaboration with Cancer Research UK.

Who it's for:
Applied clinical / biomedical researchers and health data scientists — no heavy stats background required! Attendees needed some basic knowledge of the R programming language, and were sent a small amount of pre-reading beforehand.

About the Data Diversity Theme:
Health datasets often don’t reflect the diversity of the populations they aim to serve. This lack of representativeness can limit the generalisability of research, reduce the effectiveness of new tools, and risk widening health inequalities.

A partnership between Data Science for Health Equity and Cancer Research UK, the Data Diversity Theme brings together researchers, clinicians, funders, and patient advocates to co-create practical ways of embedding diversity across the research lifecycle. It is co-led by Dr Toral Gathani (University of Oxford) and Dr Brieuc Lehmann (UCL).