Related Experiment Video
Updated: Aug 5, 2026
08:32
A Pre-Clinical Model of Synovitis Using Ex vivo Human Synovial Tissue with Preserved Function and Architecture
Published on: March 20, 2026
ChatGPT and Other Large Language Models in Inflammatory Arthritis: A Systematic Review Across Clinical Tasks
Yosef Adiniaev1, Mahmud Omar2, Tohar M Timor3
1Y. Adiniaev, Faculty of Medicine, University of Debrecen, Debrecen, Hungary; BRIDGE GenAI Lab, MA, USA.
The Journal of Rheumatology
|August 1, 2026
Summary
Large language models (LLMs) show promise for patient education and guideline adherence in inflammatory arthritis but struggle with complex clinical reasoning. Further development and prospective evaluation are needed for safe integration into rheumatology workflows.
Area of Science:
- Rheumatology
- Artificial Intelligence
- Medical Informatics
Background:
- Large language models (LLMs) are increasingly explored for healthcare applications.
- Their specific utility and limitations in managing inflammatory arthritis remain largely uncharacterized.
Purpose of the Study:
- To systematically review the existing literature on LLM performance in clinical tasks related to inflammatory arthritis.
- To identify current capabilities and gaps in LLM applications within rheumatology.
Main Methods:
- A systematic review was conducted, searching major databases (PubMed, Scopus, PubMed Central) from January 2022 to April 2026.
- 113 records were screened, with 18 studies meeting inclusion criteria for analysis.
Main Results:
- The 18 studies evaluated over 20 LLMs, predominantly GPT/ChatGPT variants, across various inflammatory arthritides.
- LLMs demonstrated utility in patient education and adherence to guidelines (48%-96% concordance), but showed lower accuracy in case-based clinical reasoning and poor agreement with real clinical data (κ ≈ 0).
- Readability assessments for LLM-generated content exceeded recommended thresholds.
Conclusions:
- LLMs can assist with patient education and factual queries under supervision but are not suitable for autonomous clinical decisions or complex reasoning in inflammatory arthritis.
- The current evidence base is nascent and predominantly based on GPT/ChatGPT models.
- Future integration requires purpose-built, knowledge-grounded systems and prospective clinical validation.
