REMOC: A CNN-Based Framework for Automated Rehabilitation Movement Classification
Fadilla ‘Atyka F. A. N. R. Nor Rashid, Ezzati E. B. Bahrom, Nurhidayah Bahar, Nor Surayahani Suriani · 2025
Physiotherapy is essential for restoring movement following illness or injury. Yet, the demand for specialists often exceeds availability, leading to challenges for patients, particularly the elderly, who struggle with regular clinic visits. To address this issue, we propose a Conventional Neural Network (CNN) framework designed to enhance physiotherapy assessments using Microsoft Kinect Xbox 360 data. In this study, we evaluated five different CNN architectures and developed an optimized framework that demonstrated superior performance in movement classification accuracy compared to existing approaches. The Rehabilitation Movement Classification (REMOC) system implements this enhanced CNN framework to facilitate home-based exercise, allowing patients to perform rehabilitation activities without frequent clinic trips. This system offers physiotherapists valuable data to assess movement correctness tailored to each patient’s needs. Our CNN framework aims to support remote physiotherapy, offering cost savings and high-quality care. Findings indicate that the proposed framework is reliable and effective for assessing physiotherapy movements in home and rehabilitation centers. This research highlights the potential of integrating technology into physiotherapy to improve patient outcomes and accessibility.