For those who would like to join the WG monthly meetings but have not put your email on the wiki page, could you do so or send email directly to me at jiang.guoqian at mayo.edu so that I can add you in the distribution list?
Has any work been done on distribution of OHDSI data and research findings through FHIR based APIs? Additionally does OHDSI currently have tools that transform FHIR based EHR extracts into the CDM format?
I am working on reading/writing/searching within OMOP thought a R4 FHIR API.
The method is quite different from the GT-FHIR one, since there is no staging area.
The OMOP table are extended a bit to deliver some FHIR elements that are not present.
The rest of the source code (based on spring boot jpa and hapi fhir) will be released and the whole will be packaged together.
As said, this is FHIR R4 on top of OMOP v6 and covers patient, encounter, condition, procedure, practitioner, practitionerRole, claim, composition, documentReference, group resources.
The architecture is based on postgresql, apache spark and apache solr. This may look quite complex to use 3 layers, but they combine greatly. Yet this FHIR API is used internally for data of 12M patients, 40M discharge summaries, 20m condition, claims, procedure and 2bilion labs with sub second responses.
The next steps include the FHIR concept/valueset/conceptmap resources to deliver the concept aspects of OMOP thought the FHIR API as referenced in this this issue. We think that FHIR value-set is an abstraction compatible with OMOP and d’like to adapt a bit the concept model to store them efficiently.