A Fast Healthcare Interoperability Resources (FHIR) layer implemented over i2b2

Area of Science:

• Health Informatics
• Computer Science
• Medical Data Management

Background:

• Electronic Health Records (EHRs) require standardized data structuring and communication.
• Current trends separate data instances from clinical information models using dual models (RM and CIM).
• HL7's Fast Health Interoperability Resources (FHIR) is a recent standard for health data exchange.

Purpose of the Study:

• To investigate the feasibility of applying the FHIR standard to model and expose EHR data.
• To integrate data from the Georges Pompidou European Hospital (HEGP) i2b2 clinical data warehouse (CDW).

Main Methods:

• Implemented a FHIR server utilizing the Java HAPI FHIR API over the i2b2 database.
• Employed a Data Access Object design pattern and FHIR resource providers.
• Tested queries for DiagnosticReport (ICD-10 codes) and MedicationOrder (French coding system) resources.

Main Results:

• Successfully exposed EHR data as FHIR resources, including DiagnosisReport, MedicationOrder, Patient, Encounter, and Medication.
• Retrieved 80 DiagnosticReport resources for 36 patients based on ICD-10 codes.
• Retrieved 503 MedicationOrder resources for 290 patients based on French medication codes.
• Validated results against ad-hoc SQL queries.

Conclusions:

• Demonstrated the feasibility of a Java layer to expose i2b2 CDW data as FHIR resources.
• Highlighted the importance of structural and semantic mapping between i2b2 and FHIR RM.
• Made source code freely available for adaptation by other i2b2 sites.