Describing patterns of diseases, medications, or other phenomena is often not sufficient to improve human health. Understanding why particular patterns exist and the potential impact of intervening to change such patterns is also often necessary. The complex scientific tasks of obtaining such an understanding can be called “causal inference”. These tasks include specifying knowledge about a system that researchers wish to study in a causal model (e.g., a causal directed acyclic graph), identifying observed data (e.g., administrative health insurance claims, electronic health records), linking the observed data to the causal model, translating the research question into target quantities, and then working to assess identifiability of those quantities, estimate them properly, and carefully interpret the estimates.

Our researchers endeavor to develop, refine, or apply new approaches that enable the successful use of data to accurately estimate disease burden; understand the effects of treatments and other exposures; and improve human health and well-being in the real world.