Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative Analysis

Abstract BackgroundRecent natural language processing breakthroughs, particularly with the emergence of large language models (LLMs), have demonstrated remarkable capabilities on general knowledge benchmarks. However, there is limited data on the performance and understanding...

Full description

Saved in:
Bibliographic Details
Main Authors: Ahmad Idrissi-Yaghir, Kamyar Arzideh, Henning Schäfer, Bahadir Eryilmaz, Mikel Bahn, Yutong Wen, Katarzyna Borys, Eva Hartmann, Cynthia Schmidt, Obioma Pelka, Johannes Haubold, Christoph M Friedrich, Felix Nensa, René Hosch
Format: Article
Language:English
Published: JMIR Publications 2025-08-01
Series:Journal of Medical Internet Research
Online Access:https://www.jmir.org/2025/1/e73540
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849223165626023936
author Ahmad Idrissi-Yaghir
Kamyar Arzideh
Henning Schäfer
Bahadir Eryilmaz
Mikel Bahn
Yutong Wen
Katarzyna Borys
Eva Hartmann
Cynthia Schmidt
Obioma Pelka
Johannes Haubold
Christoph M Friedrich
Felix Nensa
René Hosch
author_facet Ahmad Idrissi-Yaghir
Kamyar Arzideh
Henning Schäfer
Bahadir Eryilmaz
Mikel Bahn
Yutong Wen
Katarzyna Borys
Eva Hartmann
Cynthia Schmidt
Obioma Pelka
Johannes Haubold
Christoph M Friedrich
Felix Nensa
René Hosch
author_sort Ahmad Idrissi-Yaghir
collection DOAJ
description Abstract BackgroundRecent natural language processing breakthroughs, particularly with the emergence of large language models (LLMs), have demonstrated remarkable capabilities on general knowledge benchmarks. However, there is limited data on the performance and understanding of these models in relation to the Fast Healthcare Interoperability Resources (FHIR) standard. The complexity and specialized nature of FHIR present challenges for LLMs, which are typically trained on broad datasets and may have a limited understanding of the nuances required for domain-specific tasks. Improving health data interoperability can greatly benefit the use of clinical data and interaction with electronic health records. ObjectiveThis study presents the Fast Healthcare Interoperability Resources (FHIR) Workbench, a comprehensive suite of datasets designed to evaluate the ability of LLMs to understand and apply the FHIR standard. MethodsIn total, 4 evaluation datasets were created to assess the FHIR knowledge and capabilities of LLMs. These tasks include multiple-choice questions on general FHIR concepts and the FHIR Representational State Transfer (REST) application programming interface, as well as correctly identifying the resource type and generating FHIR resources from unstructured clinical patient notes. In addition, we evaluate open-source LLMs, such as Qwen 2.5 Coder and DeepSeek-V3, and commercial LLMs, including GPT-4o and Gemini 2, on these tasks in a zero-shot setting. To provide context for interpreting LLM performance, a subset of the datasets was human-evaluated by recruiting 6 participants with varying levels of FHIR expertise. ResultsOur evaluation across multiple FHIR tasks revealed nuanced performance metrics. Commercial models demonstrated exceptional capabilities, with GPT-4o achieving a 0.9990 F1 ConclusionsThis study highlights the competitive performance of both open-source models, such as Qwen and DeepSeek, and commercial models, such as GPT-4o and Gemini, in FHIR-related tasks. While open-source models are advancing rapidly, commercial models still have an advantage for specific, complex tasks. The FHIR Workbench offers a valuable platform for evaluating the capabilities of these models and promoting improvements in health data interoperability.
format Article
id doaj-art-579ffc0b53eb4f8a97d85c1b8982e055
institution Kabale University
issn 1438-8871
language English
publishDate 2025-08-01
publisher JMIR Publications
record_format Article
series Journal of Medical Internet Research
spelling doaj-art-579ffc0b53eb4f8a97d85c1b8982e0552025-08-25T20:54:01ZengJMIR PublicationsJournal of Medical Internet Research1438-88712025-08-0127e73540e7354010.2196/73540Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative AnalysisAhmad Idrissi-Yaghirhttp://orcid.org/0000-0003-1507-9690Kamyar Arzidehhttp://orcid.org/0009-0005-6074-804XHenning Schäferhttp://orcid.org/0000-0002-4123-0406Bahadir Eryilmazhttp://orcid.org/0009-0002-8743-4751Mikel Bahnhttp://orcid.org/0009-0002-0866-4023Yutong Wenhttp://orcid.org/0009-0003-8557-6665Katarzyna Boryshttp://orcid.org/0000-0001-6987-6041Eva Hartmannhttp://orcid.org/0009-0000-2600-7217Cynthia Schmidthttp://orcid.org/0000-0003-1994-0687Obioma Pelkahttp://orcid.org/0000-0001-5156-4429Johannes Hauboldhttp://orcid.org/0000-0003-4843-5911Christoph M Friedrichhttp://orcid.org/0000-0001-7906-0038Felix Nensahttp://orcid.org/0000-0002-5811-7100René Hoschhttp://orcid.org/0000-0003-1760-2342 Abstract BackgroundRecent natural language processing breakthroughs, particularly with the emergence of large language models (LLMs), have demonstrated remarkable capabilities on general knowledge benchmarks. However, there is limited data on the performance and understanding of these models in relation to the Fast Healthcare Interoperability Resources (FHIR) standard. The complexity and specialized nature of FHIR present challenges for LLMs, which are typically trained on broad datasets and may have a limited understanding of the nuances required for domain-specific tasks. Improving health data interoperability can greatly benefit the use of clinical data and interaction with electronic health records. ObjectiveThis study presents the Fast Healthcare Interoperability Resources (FHIR) Workbench, a comprehensive suite of datasets designed to evaluate the ability of LLMs to understand and apply the FHIR standard. MethodsIn total, 4 evaluation datasets were created to assess the FHIR knowledge and capabilities of LLMs. These tasks include multiple-choice questions on general FHIR concepts and the FHIR Representational State Transfer (REST) application programming interface, as well as correctly identifying the resource type and generating FHIR resources from unstructured clinical patient notes. In addition, we evaluate open-source LLMs, such as Qwen 2.5 Coder and DeepSeek-V3, and commercial LLMs, including GPT-4o and Gemini 2, on these tasks in a zero-shot setting. To provide context for interpreting LLM performance, a subset of the datasets was human-evaluated by recruiting 6 participants with varying levels of FHIR expertise. ResultsOur evaluation across multiple FHIR tasks revealed nuanced performance metrics. Commercial models demonstrated exceptional capabilities, with GPT-4o achieving a 0.9990 F1 ConclusionsThis study highlights the competitive performance of both open-source models, such as Qwen and DeepSeek, and commercial models, such as GPT-4o and Gemini, in FHIR-related tasks. While open-source models are advancing rapidly, commercial models still have an advantage for specific, complex tasks. The FHIR Workbench offers a valuable platform for evaluating the capabilities of these models and promoting improvements in health data interoperability.https://www.jmir.org/2025/1/e73540
spellingShingle Ahmad Idrissi-Yaghir
Kamyar Arzideh
Henning Schäfer
Bahadir Eryilmaz
Mikel Bahn
Yutong Wen
Katarzyna Borys
Eva Hartmann
Cynthia Schmidt
Obioma Pelka
Johannes Haubold
Christoph M Friedrich
Felix Nensa
René Hosch
Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative Analysis
Journal of Medical Internet Research
title Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative Analysis
title_full Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative Analysis
title_fullStr Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative Analysis
title_full_unstemmed Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative Analysis
title_short Using a Diverse Test Suite to Assess Large Language Models on Fast Health Care Interoperability Resources Knowledge: Comparative Analysis
title_sort using a diverse test suite to assess large language models on fast health care interoperability resources knowledge comparative analysis
url https://www.jmir.org/2025/1/e73540
work_keys_str_mv AT ahmadidrissiyaghir usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT kamyararzideh usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT henningschafer usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT bahadireryilmaz usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT mikelbahn usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT yutongwen usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT katarzynaborys usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT evahartmann usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT cynthiaschmidt usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT obiomapelka usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT johanneshaubold usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT christophmfriedrich usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT felixnensa usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis
AT renehosch usingadiversetestsuitetoassesslargelanguagemodelsonfasthealthcareinteroperabilityresourcesknowledgecomparativeanalysis