Infherno: End-to-end Agent-based FHIR Resource Synthesis from Free-form Clinical Notes
自由記述の臨床記録からFHIRリソースを合成する新アプローチ「Infherno」を発表
An agent-based system was developed to synthesize FHIR resources directly from free-form clinical notes.
Summary
A team of universities and research institutions proposed 'Infherno,' a new framework for end-to-end synthesis of FHIR resources from unstructured clinical notes. The system utilizes LLM agents and code execution to overcome limitations of previous methods, such as limited generalizability.
Details
The developed framework, 'Infherno,' aims to generate FHIR resources adhering to the HL7 FHIR standard for improving interoperability in clinical data integration and healthcare services. Previous automated translation approaches often relied on modular or LLM instruction tuning, suffering from issues like limited generalizability and structural inconsistency. To address these problems, this research proposes an end-to-end framework utilizing LLM agents, code execution, and healthcare terminology database tools. Infherno is designed to adhere to the FHIR document schema and demonstrates strong performance in predicting FHIR resources from unstructured text, even compared to a human baseline. The implementation features a front end for custom and synthetic data, supporting both local and proprietary models, thereby assisting clinical data integration processes and interoperability across institutions.
Technology Note
FHIR（Fast Healthcare Interoperability Resources）は医療データ交換の国際標準。このエントリの関連技術: HL7
Original content copyright by respective publishers