You may remember that we previously discussed FHIR and OMOP and why you need both open standards. We also talked about the collaboration between HL7 and OHDSI, the two organizations that maintain the two standards. The FHIR and OMOP harmonization project has kicked off and Evidentli is delighted to be an active supporter of this effort.
Why do FHIR and OMOP need to be harmonized?
By way of a quick reminder, FHIR is an interoperability standard that allows different healthcare systems exchange data, OMOP is a standard way to arrange health data in a database to make it easier to analyze.
The first reason that the two standards need to be harmonized is to be able to put information into systems that store data using OMOP. These databases often need to aggregate data from different systems that use FHIR. The harmonization project is developing a standard conversion from FHIR to OMOP for the benefit of analytic systems such ours.
The second reason to harmonize FHIR and OMOP is for taking information out of systems that store data in OMOP, and into, for example, AI systems. For that, the project is developing a standard conversion from OMOP to FHIR.
So where is the project up to now?
The first target is the FHIR to OMOP conversion. This has been started with mapping FHIR resources and fields into OMOP tables and columns. Some transformations are simple: A person's date of birth is a well defined concept in both models. Other transformations are not so much: FHIR records medication records by orders, administration, dispensation, prescriptions, etc. OMOP records the person's exposure to the chemical compounds, which could come from any of those. Deciding which FHIR records best represent the drug exposure, requires complex logic and more data than is available in any single FHIR message.
Project volunteers are now diligently and methodically working through hundreds of such transformation and looking to solve them. On October 15 2022, volunteers from the global FHIR and OMOP communities will gather in Bethesda, Maryland to collaborate in person. The Data Model Harmonization sub-group will have a mini-hackathon to solve some of the gnarliest of the problems identified but not solved by then. Everyone is welcome.