AI-Driven Rehabilitation Robotics: Advancements in and Impacts on Patient Recovery

摘要

The integration of artificial intelligence (AI) in rehabilitation robotics (RR) can be regarded as a significant advancement in patient recovery processes. RR refers to robotic technologies that are designed to support therapy and/or healing processes by applying tailored, adaptive, and efficient machine learning algorithms. This paper is a methodological narrative review discussing the latest trends in the application of AI in RR and its effects on the recovery of patients with stroke, spinal cord injury, and musculoskeletal disorders. Technological advancements in real-time monitoring, feedback, and intelligent data analysis have enhanced the effectiveness of these systems, providing patients with personalized rehabilitation programs that are progressive in nature, based on the patients' improvement. Research evidence demonstrates that the use of AI-driven robotic systems in clinical practice significantly improves motor skills, the rate of recovery, and overall patients' well-being, in contrast to conventional rehabilitation procedures. Moreover, the review highlights the drawbacks and problems that present-day technologies have; these problems concern the costs, the absence of unified protocols, and legal requirements. Furthermore, the combination of AI with RR will likely progress even more, with ongoing research studying ways in which the system's intelligence could be upgraded, patients' involvement in the process could be increased, and the applicability of these technologies to various forms of needs could be increased as well. The present review thus highlights the importance and the role of AI-RR in enhancing the patient's recovery and opens the way for therapeutic practices. This narrative review was informed by targeted searches in PubMed, Scopus, and IEEE Xplore covering 2015-2025, with an emphasis on English-language, peer‑reviewed studies in stroke, spinal cord injury, and musculoskeletal rehabilitation; no formal meta-analytic synthesis was undertaken, consistent with a narrative approach.