As the availability of high-quality, routinely collected health data continues to grow, the ability to draw credible causal conclusions from real-world data has become a critical skill for modern health data scientists. The Causal Inference module introduces students to the principles, assumptions, and practical tools required to move beyond association and towards answering meaningful causal questions in healthcare and public health.
This module provides rigorous yet accessible training in causal inference methodology, covering the potential outcomes framework for expressing causal effects and the key assumptions required to support causal claims. You will learn how to use and interpret Directed Acyclic Graphs (DAGs) to reason about confounding, bias, and study design, and how causal thinking underpins the analysis of randomised clinical trials.
Building on this foundation, the module explores advanced topics including non-compliance and Complier Average Causal Effects, as well as a range of widely used causal methods for observational data. These include G-computation, propensity score methods, difference-in-differences, and synthetic control approaches equipping you with a versatile toolkit for evaluating interventions when randomisation is not feasible.
A strong emphasis is placed on applied learning. Through hands-on guided practical sessions, you will gain experience implementing causal methods using real-world datasets drawn from clinically important settings such as aneurysm repair, maternal health, and cancer research. These sessions focus not only on estimation, but also on interpretation, transparency, and the limitations of causal analyses in practice.
By the end of the module, you will be able to critically evaluate causal claims in the health literature, design analyses that align with clear causal questions, and confidently apply modern causal inference methods to real-world health data.
The module is delivered over 12 weeks, consisting of 12 two-hour lectures and 12 two-hour practical sessions, combining conceptual understanding with applied data analysis. It provides essential training for students aiming to work in health data science, epidemiology, clinical research, and policy evaluation, where robust causal reasoning is central to evidence-based decision-making.