Creative Commons source reading · Rehabilitation and therapy
ChatGPT-generated rehabilitation programs in sports physiotherapy: an expert evaluation and a mixed-methods study of clinical applicability
Cali A; Yorukoglu ME; Acar G; Safran E
Originally published in Front Med (Lausanne) · 2026-08-26
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
AbstractBackgroundIndividualized rehabilitation supports safe return to sport, yet clinical workload may limit personalized program design. This study evaluated ChatGPT-4.1-generated programs for five sports injury scenarios using a structured rubric scored by two independent sports physiotherapy experts and examined inter-rater reliability and clinical applicability.MethodsMixed-methods design combining rubric scoring with deductive qualitative content analysis of expert commentary. Five cases covered muscle (hamstring strain), ligament (anterior cruciate ligament [ACL] reconstruction), tendon (rotator cuff tendinopathy), neural (lumbar disc herniation) and bone (clavicle fracture) injuries. Exactly one 6-week tabular program per case was generated with ChatGPT-4.1 from a standardized prompt in a fresh session with memory, cross-chat…
Discussion
This study systematically examined the quality and clinical applicability of rehabilitation programs generated by ChatGPT-4.1 for five different sports injury scenarios through evaluation by two independent experts. The overall mean score (3.85/5) indicated moderate-to-good performance by ChatGPT-4.1, but considerable variability was observed across cases (ranging from 1.88 to 5.00). This finding is consistent with previous literature indicating that LLM performance varies substantially according to the complexity of the clinical scenario (7, 8). The most striking finding is the performance gap between cases based on linear, largely standardized protocols (clavicle fracture rehabilitation, Case 5) and cases requiring multifactorial, postoperative-staging-sensitive, individualized risk assessment (ACL reconstruction, Case 2). For the…
Conclusion
This mixed-methods study demonstrated that ChatGPT-4.1 can generate structured rehabilitation programs of moderate-to-good quality (overall mean 3.85/5, 95% CI 3.48–4.22) for five different sports-related clinical scenarios, with good inter-rater reliability (ICC = 0.84, 95% CI 0.52–0.94). Performance varied with case complexity: the highest ratings were obtained for the linear, protocol-based scenario (clavicle fracture rehabilitation after union) and the lowest for the scenario requiring multifactorial postoperative individualization (ACL reconstruction). Because each tissue category is represented by a single scenario (n = 1), these are observations about five specific generated programs rather than established properties of the model, and the direction of the difference — not its magnitude — is what the present data support. The…
Editorial reading guide
- How were injury, recovery and exposure defined and verified?
- Which criteria support progression, and what finding should stop or change the plan?
