AI in Rehabilitation: Evidence-Based Update

Course overview

Artificial intelligence is fundamentally reshaping the practice of rehabilitation medicine, introducing transformative capabilities in patient assessment, clinical decision-making, intervention delivery, and outcome prediction that span the entire continuum of care. For physical therapists, occupational therapists, and nurses working across inpatient, outpatient, and home-based settings, these technologies represent both an unprecedented opportunity to enhance clinical effectiveness and a significant professional challenge requiring new competencies, workflows, and ethical frameworks. Understanding the evidence base supporting AI-driven rehabilitation tools is now essential for clinicians who seek to deliver care that reflects the current standard of practice and the evolving expectations of patients, payers, and accreditation bodies. A living systematic mapping review of 240 studies published in the European Journal of Physical and Rehabilitation Medicine confirmed that AI has been tested across all stages of the rehabilitation process, with intervention representing 23.8% of applications, prognosis 17.5%, assessment 16.7%, diagnosis 12.9%, and monitoring 12.5%. The majority of these studies focused on neurological conditions (57.9%) and orthopedic rehabilitation (22.7%), with stroke, Parkinson's disease, and amputation emerging as the most commonly studied diagnoses. The breadth of AI tools now available for rehabilitation practice is captured comprehensively in the JAMA Summit Report on Artificial Intelligence, which categorizes AI tools influencing health care delivery into distinct functional domains spanning clinical decision support, predictive analytics, workflow automation, and patient-facing applications. These domains map directly onto the daily practice of rehabilitation professionals, from sensor-based gait analysis and robotic-assisted therapy to AI-powered documentation systems and predictive discharge planning. Despite this technological momentum, significant implementation challenges persist that must be honestly confronted if AI integration is to benefit rather than burden clinical practice. Over half of published AI rehabilitation studies lack a comparator group (50.8%), external validation has been applied in only 5.8% of studies, and explainability has been incorporated in just 10.2%. Most AI applications rely on supervised learning (70.8%) with small, single-type datasets, limiting generalizability to diverse patient populations and real-world clinical environments. Furthermore, adoption rates vary substantially across rehabilitation disciplines: 63.3% of physical therapists report no experience with AI applications at work, occupational therapists face significant institutional barriers to implementation, and only 13.8% of clinicians overall feel that their training adequately prepared them for AI integration. These gaps between technological capability and clinical readiness represent both a challenge and an urgent opportunity for the rehabilitation workforce. This course provides an evidence-based examination of AI adoption in rehabilitation, synthesizing current research across assessment technologies, clinical effectiveness, patient population considerations, discipline-specific applications, workforce readiness, clinical documentation, and implementation strategy. The content draws on systematic reviews, meta-analyses, and randomized controlled trials published predominantly between 2022 and 2025, with an emphasis on quantifiable outcomes and actionable clinical implications. Throughout the course, attention is given to the unique roles and perspectives of physical therapists, occupational therapists, and nurses, and to the ethical, equity, and implementation considerations that must inform responsible AI integration.