Adherence Monitoring and Improvement in Physiotherapy Rehabilitation: A Narrative Review of Artificial Intelligence and Wearable Technology Application

Abstract

Background: Patient adherence to physiotherapy rehabilitation protocols remains a critical challenge globally, with rates as low as 40–65% reported in neurological and musculoskeletal populations. Artificial intelligence (AI) and wearable technology offer promising solutions for objective monitoring and improving exercise adherence beyond the clinic.
Objectives: To review the current evidence on AI and wearable technology applications for monitoring and improving patient compliance with physiotherapy rehabilitation programmes.
Methods: A narrative review methodology was employed consistent with PRISMA 2020 reporting guidelines and structured using the PICO framework. Five electronic databases were searched (PubMed, Scopus, Cochrane Library, PEDro, and Web of Science) for studies published between January 2018 and April 2026 using search terms related to artificial intelligence, machine learning, wearable technology, and physiotherapy adherence. Studies involving any adult demographic and any physiotherapy environment were incorporated.
Results: Six core AI mechanisms were identified across the thirteen included studies: real-time monitoring, personalised feedback, gamification, predictive non-adherence detection, automated reminders and adaptive exercise progression. Six of the thirteen included studies addressed predictive non-adherence detection and adaptive exercise progression which were the most widely supported mechanisms. In contrast, five studies supported real-time monitoring, gamification, and automated reminders. Wearable smart watches and IMU-based systems provided strong and consistent evidence for objective real-time exercise recognition in both clinical and home settings. AI-driven virtual assistants and gamification platforms demonstrated the greatest potential to improve patient engagement and motivation in home-based rehabilitation. Evidence for gamification and adaptive progression was primarily derived from review-level and observational studies rather than from primary, randomised experimental research, leading to a need for robust clinical trials in these areas.
Conclusion: AI and wearable technology offer a transformative but underused opportunity for monitoring adherence to physiotherapy. Robust clinical trials, particularly in neurological rehabilitation, are urgently needed. Physiotherapists and researchers must engage with these technologies to ensure evidence-based, equitable, and patient-centred implementation.