The Role of Artificial Intelligence in Treatment Planning in Physical and Rehabilitation Medicine

OBJECTIVE: Individualized rehabilitation plans are crucial for maximizing clinical outcomes. Generative artificial intelligence (AI), particularly large language models (LLMs), which can generate new content based on large knowledge bases, could improve personalized rehabilitation. This systematic review aimed to assess accuracy and application of LLMs in rehabilitation. DESIGN: A systematic search was conducted in PubMed, Web of Science, and Scopus until February 2026. The PICO included LLMs (P), their use in rehabilitation planning (I), comparison with expert opinion or other AI systems (e.g., machine learning, deep learning, or CDSS) (C), the ability to generate appropriate rehabilitation plans and compliance with expert clinicians or clinical practice guidelines (O). Studies involving non-generative AI were excluded. Methodological quality was assessed with JBI tools. Data extraction was performed by two reviewers; a third resolved conflicts. RESULTS: Nine studies met inclusion criteria. Concordance with expert opinion varied, reaching 98.75% after an expert advisory board reviewed AI-generated statements. In musculoskeletal and cardiovascular rehabilitation, LLMs demonstrated alignment with guidelines. However, the included studies reported hallucinations, underscoring the necessity of supervision. CONCLUSION: LLMs showed promising capabilities in supporting rehabilitation treatment planning, possibly offering adaptability and personalization. Nonetheless, high-quality studies are necessary to validate real-world effectiveness and define implementation strategies.