Executive summary
Unless physiotherapists are actively engaged in a conversation around AI in clinical practice we run the risk that our clinical decision-making will be subject to machine intelligence, rather than informed by it.
Rowe, Nicholls & Shaw (2022)
This discussion document explores the integration of generative artificial intelligence (generative AI) in physiotherapy, addressing its impact on clinical practice, education, and research. It aims to provide physiotherapy professionals, faculty members, researchers, students, and other stakeholders, with a comprehensive overview of the current capabilities, opportunities, and challenges, presented by this rapidly evolving technology. The document emerged out of planning for the 2024 IFOMPT conference and was informed by a series of activities that took place before, during, and after the event.
The document begins by introducing key concepts of generative AI, emphasising its role as a powerful assistant that can support human capabilities across a range of contexts relevant to physiotherapy practice. It outlines current applications across different areas of the profession, while acknowledging that these applications will likely evolve rapidly. With that in mind, we also try to avoid making any speculative claims about what AI might do in the future, and ensure that all claims linked to model capability are anchored to what is possible today, or at least to something clearly visible on the horizon.
We believe there are significant opportunities for enhancing physiotherapy practice through generative AI, including improved clinical reasoning and treatment planning, personalised patient care, enhanced professional education, real-time feedback and monitoring, administrative efficiency, enhanced patient education, and expanded research capabilities.
However, the document also addresses a wide range of potential risks and challenges. These include concerns about deskilling and over-reliance on AI, data privacy and security issues, errors and unclear liability, the potential for bias, depersonalisation of care, lack of transparency when using AI, and ethical and regulatory challenges. In addition, the environmental and economic sustainability of generative AI systems, given their high energy demands and development costs, should be considered when evaluating their long-term role in physiotherapy (however, we have chosen not to include a comprehensive overview of sustainability concerns in the first version of the document).
To maximise opportunities while mitigating risks, the document proposes several strategies in each section that we suggest are the focus of this discussion document. In other words, the aim of the document is to inform and provide a space for this community to explore our options for moving forward. These strategies include the development of comprehensive AI literacy programmes for all stakeholders, development of ethical guidelines and best practices, emphasis on human-AI collaboration, ongoing research and evaluation, adoption of transparent and explainable AI systems, adaptive regulation, and active stakeholder engagement.
Ultimately, this document serves as a starting point for critical discussions about the future of generative AI in physiotherapy. It encourages all professional stakeholders to actively engage with these technologies, think critically about their implementation and evaluation, and contribute to shaping the responsible development and use of AI in the field.
Note: This document is still in active development and includes unsubstantiated claims (although supporting evidence is being regularly added) and minor editorial issues. You could consider it a version 0.7 (as of 30 May 2025). However, the general outline is unlikely to change much.
Last updated