CHSPR Seminar | Boosting Real-World Evidence in Health Services

Using Machine Learning and Robust Methods to Harness High-Dimensional Proxies from Administrative Data for Minimizing Residual Confounding

Ehsan Karim, UBC School of Population and Public Health

Thursday Sept 25, 2025
12-1 pm PT
SPPH B104 or Zoom (Ehsan Karim will speak in person)
Register

This seminar is a health services methods talk and focuses on how to utilize health services data better. The talk explores the effectiveness of machine learning extensions and doubly robust estimators in integrating high-dimensional proxies from administrative health data to minimize residual confounding in observational studies. Drawing from simulations based on real-world data, the talk highlights how incorporating proxies substantially reduces bias and improves coverage compared to methods without them, while emphasizing the trade-offs of complex learner libraries. Attendees will gain insights for enhancing real-world evidence in health services research, including tailored strategies for robust causal inference in pharmacoepidemiology and chronic disease management.

Dr Ehsan Karim is an Associate Professor in Health Data Science at the UBC School of Population and Public Health, a Scientist at the Centre for Advancing Health Outcomes, and a specialist in causal inference, machine learning, biostatistics, and pharmacoepidemiology. His research focuses on developing methodologies for analyzing observational data in large healthcare datasets, with applications in comparative effectiveness research.

Register in advance for this seminar. After registering, you will receive a confirmation email containing information about joining the meeting.