Spondyloarthritis & psoriatic arthritis
ChatGPT vs rheumatologists: cross-sectional study on accuracy and patient perception of AI-generated information for psoriatic arthritis
Authors' conclusion
In this exploratory study, ChatGPT-4 appeared able to generate accurate and readable responses to PsA-related questions and was often preferred by patients.
Study context
- Population
- Psoriatic arthritis (PsA) patients generating questions (32 questions, 6 themes); answers scored by 14 clinicians; separate arm of 67 PsA patients rating preference67 patients in preference arm
- Participants
- 76 patients
- Exposure
- ChatGPT-4 answers vs answers by Italian PsA specialists
- Expertsn = 12reference
- Outcome
- Accuracy (1-5 Likert scale) and completeness (1-3) of answers, scored by clinicians
Methodological detailsexploratory, descriptive comparison, mixed-effects ordinal logistic models, no formal sample size calculation
- Exploratory, descriptive comparison
- Mixed-effects ordinal logistic models
- No formal sample size calculation
Main finding
As reported in the abstract
- Accuracy appeared comparable between ChatGPT and experts
- ChatGPT scored lower in completeness
- Accuracy lower for pregnancy/fertility, no clear relevant differences in other domains
- P values descriptive
Observational data: associations, not effects.
Further findings
| Outcome | Experts | P value |
|---|---|---|
| Patient preference (998 choices)ChatGPT vs Experts · no preference 164/998 (16.4%) | 34.4% | P < 0.001 |
- Patient preference (998 choices): ChatGPT 49.2%
Patient preference (998 choices)
- Relative preference for ChatGPT responses in prognosis and therapy
Readability
- ChatGPT responses on average more readable across indices
Information most sought by patients
- Prognosis/comorbidities (54/76)
- 71.1%
- Therapy strategy (48/76)
- 63.2%
- Treatment risks (38/76)
- 50.0%
Original publication
ChatGPT vs rheumatologists: cross-sectional study on accuracy and patient perception of AI-generated information for psoriatic arthritis. Ann Rheum Dis. Published 12 December 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →