Large Language Models for Automating Conformance to Health-Data Standards: The Interoperability Case of HL7 FHIR and...

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare
• Clinical Data Management

Background:

• Healthcare systems rely on interoperability standards for seamless data exchange and secondary data analysis.
• Transforming diverse clinical data into standards-compliant formats is complex, labor-intensive, and prone to errors.
• Steep learning curves and technical requirements limit practitioner access to these standards.

Purpose of the Study:

• To investigate the potential of large language models (LLMs) to automate the conformance of clinical data to interoperability standards.
• To assess if LLMs can maintain syntactic validity, semantic correctness, and reproducibility during data transformation.
• To simplify the adoption and utilization of healthcare data standards for practitioners.

Main Methods:

• Developed and validated a prototype system using large language models.
• Utilized diabetes measurements as a specific domain for testing.
• Evaluated extraction fidelity, coding accuracy against controlled vocabularies, and conformance to key standards (e.g., FHIR, OMOP).

Main Results:

• The LLM-based prototype successfully automated the conversion of clinical data to standards-compliant formats.
• Demonstrated preservation of syntactic validity, semantic correctness, and reproducibility.
• The system accurately extracted and coded diabetes measurements, emitting FHIR Observations and OMOP MEASUREMENT rows.

Conclusions:

• Large language models show significant promise in automating the transformation of clinical data to interoperability standards.
• LLMs can simplify data management, enhance data accessibility for secondary uses, and reduce the burden on healthcare practitioners.
• This approach facilitates wider adoption and effective utilization of healthcare data standards, improving data analysis capabilities.