TeleRehab AI: An AI-Powered Mobile Physiotherapy Assistant

This paper introduces TeleRehab AI, an intelligent mobile-based physiotherapy system designed to assist patients in performing rehabilitation exercises independently. The system combines conversational AI with real-time pose analysis to deliver personalized and interactive guidance. It integrates four major components: a fine-tuned LLaMA 3.2 3B language model for physiotherapy consultation, a structured multi-layer message processing pipeline, an on-device pose estimation module using Google ML Kit BlazePose, and a lightweight exercise classification model deployed via TensorFlow Lite. The system evaluates squat performance using biomechanical parameters such as knee angle, torso inclination, and movement depth. A Support Vector Machine model achieves an accuracy of 84.8%, and is further compressed into a compact neural network using model distillation, achieving 99.7% agreement while maintaining a model size of only 4.9 KB. Unlike traditional solutions, the proposed system operates directly on mobile devices without continuous internet access, making it suitable for large-scale deployment in resource-constrained environments.

Telerehabilitation, pose estimation, LLM fine-tuning, on-device inference, TFLite, exercise classification, physiotherapy AI, mobile health, biomechanical analysis

1. M. A. Ashraf, S. Najam, T. Sadiq, S. Algamdi, H. Aljuaid, H. Rahman, and A. Jalal, “A Novel Telerehabilitation System for Physical Exercise Monitoring in Elderly Healthcare,” IEEE Access, 2025. [Google Scholar] [Crossref]

2. W. I. W. Wan Noor Asmawi and S. Prabha Kumaresan, “Real-time posture monitoring for effective exercise using MediaPipe Python,” Journal of Informatics and Web Engineering, vol. 5, no. 1, Feb. 2026, eISSN: 2821-370X. [Google Scholar] [Crossref]

3. W. Simoes, L. Reis, C. Araujo, and J. Maia Jr., “Accuracy Assessment ˜ of 2D Pose Estimation with MediaPipe for Physiotherapy Exercises,” in Proc. 14th Int. Conf. Current and Future Trends Information and Communication Technologies in Healthcare (ICTH), 2024. [Google Scholar] [Crossref]

4. V. Wagh, M. W. Scott, and S. N. Kraeutner, “Quantifying Similarities Between MediaPipe and a Known Standard to Address Issues in Tracking 2D Upper Limb Trajectories: Proof of Concept Study,” JMIR Formative Research, 2024. [Google Scholar] [Crossref]

5. A. Jaiswal, G. Chauhan, and N. Srivastava, “Using learnable physics for real-time exercise form recommendations,” arXiv preprint arXiv:2301.xxxx, 2023. [ [Google Scholar] [Crossref]

6. 6] Y. Li, J. H. Choi, J. Cheng, A. F. Martins, and G. Krueger, “ChatDoctor: A Medical Chat Model Fine-Tuned on LLaMA Model Using Medical Domain Knowledge,” arXiv preprint arXiv:2303.14070, 2023. [Google Scholar] [Crossref]

7. T. Han, L. C. Adams, J. M. Papaioannou, P. Nachev, and S. Ourselin, “MedAlpaca—An Open-Source Collection of Medical Conversational AI Models and Training Data,” arXiv preprint arXiv:2304.08247, 2023. [Google Scholar] [Crossref]

8. A. Jaiswal, G. Chauhan, and N. Srivastava, “Using Learnable Physics for Real-Time Exercise Form Recommendations,” arXiv preprint arXiv:2301.xxxx, 2023. [Google Scholar] [Crossref]

9. C. Mennella, U. Maniscalco, G. De Pietro, and M. Esposito, “A Deep Learning System to Monitor and Assess Rehabilitation [Google Scholar] [Crossref]

10. Exercises in Home-Based Remote and Unsupervised Conditions,” Computers in Biology and Medicine, vol. 163, p. 107,175, 2023. [Google Scholar] [Crossref]

11. E. J. Hu, Y. Shen, P. Wallis, Z. Allen-Zhu, Y. Li, S. Wang, L. Wang, and W. Chen, “LoRA: Low-Rank Adaptation of Large Language Models,” in Proc. Int. Conf. Learning Representations (ICLR), 2022. [Google Scholar] [Crossref]

12. N. Nakano, T. Sakura, K. Ueda, L. Omura, A. Kimura, Y. Iino, S. Fukashiro, and S. Yoshioka, “Evaluation of 3D Markerless Motion Capture Accuracy Using OpenPose With Multiple Video Cameras,” Frontiers in Sports and Active Living, vol. 2, p. 50, 2020. [Google Scholar] [Crossref]

13. V. Bazarevsky, I. Grishchenko, K. Raveendran, T. Zhu, F. Zhang, and M. Grundmann, “BlazePose: On-Device Real-Time Body Pose Tracking,” arXiv preprint arXiv:2006.10204, 2020. [Google Scholar] [Crossref]

14. Y. Gu, S. Pandit, E. Saraee, T. Nordahl, T. Ellis, and M. Betke, “HomeBased Physical Therapy with an Interactive Computer Vision System,” in Proc. IEEE/CVF Int. Conf. Computer Vision Workshops (ICCVW), 2019. [Google Scholar] [Crossref]

15. A. Rintala, S. Yalen, J. Paltamaa, A. Heinonen, J. Karvanen, and [Google Scholar] [Crossref]

16. ´ M. Arkela-Kautiainen, “Effectiveness of Technology-Based Distance Interventions for Older People: A Systematic Review,” Journal of Telemedicine and Telecare, vol. 25, no. 4, pp. 205–220, 2019. [Google Scholar] [Crossref]

17. M. Abadi, A. Agarwal, P. Barham, E. Brevdo, Z. Chen, C. Citro, et al., “TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems,” in Proc. 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI 16), pp. 265–283, 2016. [Google Scholar] [Crossref]

18. G. Hinton, O. Vinyals, and J. Dean, “Distilling the Knowledge in a Neural Network,” in NeurIPS Deep Learning Workshop, 2015. [Google Scholar] [Crossref]

19. M. Tousignant, L. Brosseau, B. C. Craven, and K. A. Lowry, “Supervised Exercise Rehabilitation After Orthopedic Surgery: A Randomized Controlled Trial,” Arthritis Care & Research, vol. 63, no. 8, pp. 1108– 1116, 2011. [Google Scholar] [Crossref]

20. J. Platt, “Probabilistic Outputs for Support Vector Machines and Comparisons to Regularized Likelihood Methods,” in Advances in Large Margin Classifiers, MIT Press, 1999, pp. 61–74. [Google Scholar] [Crossref]

• Block-Based Programming for Education: A Comprehensive Analysis of Visual Programming Environments in K-12 Learning
• Management of Academic Libraries and Client Satisfaction Towards Digital Utilization: Basis for Monitoring Library Operations in SOCCSKSARGEN Region.
• Revenue Leakages in TPA Insurance Claims and Corporate Claims: An Institutional Overview of Aster Prime Hospital, Hyderabad
• Technology and Innovation in Hospitality and Tourism: A Management Perspective
• Geospatial Distribution of Tarok Sacred Grove of Langtang North and Langtang South Local Government Areas