Task: Implement analytic methods to identify associations and measuring the performance characteristics of identification methods
Various stakeholders have proposed alternative solutions for analyzing observational data to identify drug-condition associations. These methods are applied to ‘screen’ for associations between drugs and events of interest (designated medical events / health outcomes of interest) or general data mining for events that are not specified a priori. OMOP intends to be inclusive in its objective tests of method performance by starting with a large pool of methods and attempting to engage the originators of those methods in the programming and implementation of specific tests against the core data sources where possible.
Feasibility. The initial agreed pool of potential methods to be studied is listed in the “Methods Matrix." The Research Core will apply each method to the databases using the common data model and diagnostic tools to determine the feasibility of implementing these methods in the databases (do they actually run in a reasonable period of time).
The Extended Consortium will be encouraged to conduct similar analyses on their data sources using the protocols and tools made available by the Research Core and publish their findings .
Deliverable: OMOP will produce an open source repository of methods and code for the method implementations that can be applied to data sources normalized to the OMOP common data model.