Smart Recovery 2.0: A Markerless Computer Vision Platform for Gamified Neuro-Rehabilitation

Abstract

This paper presents Smart Recovery 2.0, a clinician-guided hybrid telerehabilitation platform integrating edge-computed artificial intelligence with direct medical oversight in a browser-native environment. This work extends the previously published Smart Recovery system (IRE Journal) through expanded rehabilitation modules, AI-assisted dashboard features, and a patient verification mechanism. To address scalability and cost limitations in conventional neuro-rehabilitation, the platform delivers 60 game modules spanning six anatomical domains: face, fingers, wrist, hands, arms, and shoulders. A secure remote configuration pipeline links the doctor dashboard with the patient application, enabling clinicians to review movement data and prescribe customized therapy sessions. System tracking and adherence were evaluated using a Controlled Healthy-Subject Verification Protocol (CHSVP). The markerless motion extraction framework achieved a spatial error of Edist =3.12±0.45 pixels with an end-to-end processing latency of ≤65 ms.