Automating FHIR Element Recognition to Enhance Semantic Mapping in Oncology Informatics

In health IT software development, managing Health Level Seven Fast Healthcare Interoperability Resources (HL7 FHIR) data model is crucial for interoperability. However, manual identification of the FHIR resource and element corresponding to a concept in a site electronic health record can be cumbersome and error-prone. Literature describing correct use of FHIR identifiers is sparse, forcing informaticists to adapt a data management strategy for HL7 FHIR requirements, company database conditions and departmental workflows. Methods: We embarked on a quality improvement initiative to enhance FHIR resource management at a clinical decision support software development company. The project's objectives were: (1) building a Device profile for the company's FHIR model, (2) developing the Structured Query Language (SQL) query for retrieving FHIR Uniform Resource Identifiers (URIs), (3) enhancing the concept map user interface (UI), and (4) preventing data entry errors in FHIR URI creation. Interventions included profile review, query testing, user interface enhancements, and workflow analysis. User acceptance testing and click count reduction were used for evaluation. Results: All project objectives were achieved, with iterative improvements obtained through adherence to Agile methodology. The Device profile yielded no critical errors, and provided the context for development of the FHIR URI query. Large language model (LLM) prompting was used to develop the query. User interface improvements were made to the internal tools used by company informaticists, including a display for the FHIR URI and a structured form for creating FHIR URIs. User feedback and workflow analysis for both interventions were positive. Conclusions: Stakeholder collaboration, thorough documentation and iterative refinement were pivotal in achieving project objectives. Recommendations include extending database integrity practices and automating concept map registration to benefit the informatics community.