Evaluating ChatGPT’s Role in Musculoskeletal Rehabilitation

AI tools like ChatGPT offer promising support in physiotherapy by providing rapid clinical insights. However, before such tools can be meaningfully integrated into musculoskeletal rehabilitation, their alignment with evidence-based clinical practice guidelines (CPGs) must be critically evaluated.

A study from Turkey assessed the performance of ChatGPT in responding to physiotherapy-related queries by comparing its responses against established CPGs in the domains of disease information, patient assessment, and rehabilitation.

Methods
Twenty clinical questions were developed by two experienced musculoskeletal physiotherapists, covering the upper extremity (7), lower extremity (9), and spine (4). Questions spanned three domains: disease information, assessment, and rehabilitation. ChatGPT’s responses were scored independently by two raters using a 5-point Likert scale to evaluate relevance, accuracy, clarity, completeness, and consistency.

Key Findings

• High Performance in Clarity and Relevance: ChatGPT produced well-structured, understandable responses with high average scores in clarity (4.85/5) and relevance (4.50/5).
• Strong Alignment with Guidelines in Disease Information: The AI showed the most consistency and accuracy when answering questions related to diagnosis or general disease knowledge.
• Weaker in Rehabilitation Guidance: Responses related to treatment interventions and rehabilitation planning were less consistent (average consistency score: 3.85), highlighting variability in ChatGPT’s answers across repeated questions.

Conclusions
ChatGPT shows strong potential as a supplementary tool for physiotherapists, particularly in delivering accurate and clearly structured disease-related information. However, its lower performance in rehabilitation and variability in consistency suggest that it should not be used as a standalone clinical decision-making tool. Its limitations in nuanced reasoning and response stability emphasize the need for further refinement, especially for domain-specific applications in physiotherapy.

Clinical Implications

• Use ChatGPT as a quick-reference resource, not as a primary clinical advisor.
• Exercise professional judgment when interpreting AI-generated recommendations, particularly for individualized rehabilitation planning.
• Consider its role in clinical education and early-stage information retrieval, while promoting critical engagement with guidelines and clinical reasoning.