ARTIFICIAL INTELLIGENCE IN CLINICAL REHABILITATION: A SYSTEMATIC REVIEW OF CLINICAL APPLICATIONS, OUTCOMES AND IMPLEMENTATION CHALLENGES

Abstract

Artificial intelligence (AI) is increasingly being incorporated into clinical rehabilitation through computer vision, wearable sensors, machine learning, robotics, virtual reality, mobile applications, and remote monitoring. This structured evidence review summarizes published evidence on the clinical applications, outcomes, and implementation challenges of AI-enabled rehabilitation. Evidence was synthesized from systematic reviews, scoping reviews, mapping studies, and selected clinical studies. Reported applications include movement assessment, gait analysis, stroke rehabilitation, musculoskeletal rehabilitation, prediction of functional outcomes, personalized exercise prescription, telerehabilitation, and robotic assistance. The literature indicates substantial potential for improving assessment precision, monitoring, personalization, and access to rehabilitation. However, the evidence remains heterogeneous, with frequent limitations related to small datasets, inadequate external validation, limited explainability, equipment cost, privacy, workflow integration, and insufficient long-term clinical evaluation. AI should therefore be considered a clinical decision-support and rehabilitation-enhancement technology rather than a replacement for professional physiotherapy judgment.