Performance and Practical Considerations of Large and Small Language Models in Clinical Decision Support in Rheumatology

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Felde, Sabine, Buchkremer, Rüdiger, Chehab, Gamal, Thielscher, Christian, Distler, Jörg HW, Schneider, Matthias, Richter, Jutta G.
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866916837443239936
author Felde, Sabine
Buchkremer, Rüdiger
Chehab, Gamal
Thielscher, Christian
Distler, Jörg HW
Schneider, Matthias
Richter, Jutta G.
author_facet Felde, Sabine
Buchkremer, Rüdiger
Chehab, Gamal
Thielscher, Christian
Distler, Jörg HW
Schneider, Matthias
Richter, Jutta G.
contents Large language models (LLMs) show promise for supporting clinical decision-making in complex fields such as rheumatology. Our evaluation shows that smaller language models (SLMs), combined with retrieval-augmented generation (RAG), achieve higher diagnostic and therapeutic performance than larger models, while requiring substantially less energy and enabling cost-efficient, local deployment. These features are attractive for resource-limited healthcare. However, expert oversight remains essential, as no model consistently reached specialist-level accuracy in rheumatology.
format Preprint
id arxiv_https___arxiv_org_abs_2507_07983
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Performance and Practical Considerations of Large and Small Language Models in Clinical Decision Support in Rheumatology
Felde, Sabine
Buchkremer, Rüdiger
Chehab, Gamal
Thielscher, Christian
Distler, Jörg HW
Schneider, Matthias
Richter, Jutta G.
Computation and Language
Artificial Intelligence
L01.224.900.500 (Primary), L01.700.508.300, L01.224.050.375, H02.403.720.750, N04.590, N04.452.758.625 (Secondary)
I.2.7; H.3.3; J.3; I.2.9; C.4
Large language models (LLMs) show promise for supporting clinical decision-making in complex fields such as rheumatology. Our evaluation shows that smaller language models (SLMs), combined with retrieval-augmented generation (RAG), achieve higher diagnostic and therapeutic performance than larger models, while requiring substantially less energy and enabling cost-efficient, local deployment. These features are attractive for resource-limited healthcare. However, expert oversight remains essential, as no model consistently reached specialist-level accuracy in rheumatology.
title Performance and Practical Considerations of Large and Small Language Models in Clinical Decision Support in Rheumatology
topic Computation and Language
Artificial Intelligence
L01.224.900.500 (Primary), L01.700.508.300, L01.224.050.375, H02.403.720.750, N04.590, N04.452.758.625 (Secondary)
I.2.7; H.3.3; J.3; I.2.9; C.4
url https://arxiv.org/abs/2507.07983