Developing a structured process for measuring and interpreting health outcomes of interest in the OMOP common data model

The first two years of OMOP research yielded significant progress towards structured use of existing observational data sources for active surveillance. OMOP established distributed and centralized data access mechanisms with claims and EHR data transformed to a common data model. In 2011, one of OMOP's priorities include refinement of some of the strategies previously developed. In particular, definitions for measurement of health outcomes of interest (HOI) in observational data are a critical element to bolster confidence in output of the active surveillance process.

Our research will investigate 3-4 of the OMOP HOIs (and method data management strategies), starting with acute liver injury, and their impact on method performance. The overall goal of this work is to minimize false positive cases through better measurement of the HOI. Our process includes refining HOI definitions through clinical review of cases, creation of training datasets through expert classification of true cases, predictive modeling to further refine HOI definitions, and comparison of methodological performance based on probability thresholds.

Deliverables of this research will be posted below as available.