GPT-4o recognizes osteoarthritis from patient questionnaires in a selected cohort
Source check: Source checked. Not independent scientific confirmation.
Finding: Researchers compared model-generated diagnoses from patient questionnaires with blinded clinician assessments. The primary results table records 85 correct GPT-4o diagnoses among 93 selected osteoarthritis cases. The study excluded non-osteoarthritis diagnoses, so it does not establish specificity, general triage accuracy or clinical benefit.
Technology: GPT-4o questionnaire-based diagnostic generation
Family GPT-4: profile · models page
Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.
- AI contribution
- GPT-4o inferred diagnoses from patient-reported symptoms and history; its output was compared with independent clinician assessment.
- Test and result
- 115 questionnaires collected prospectively;93 clinician-confirmed osteoarthritis cases analyzed after exclusions. Clinicians were blinded to questionnaires; deidentified responses were subsequently submitted to pretrained models.
- Limitation / still unknown
- Source headline sensitivity 92.3% conflicts with Table 2 count85/93 (91.4%); represented result follows the explicit count. Negative diagnoses were excluded and one entirely symptom-incomplete questionnaire removed. GPT-4o helped design the questionnaire; prompts/repeated-generation details in supplements were uninspected. No locked deployment or outcome benefit.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC11754967/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Actual primary cohort/model Methods, Results/Table2 and limitations in supported Chrome; source-front onlineDecember26,2024 versus2025collection distinguished. Internal headline/count discrepancy retained..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.