Hybrid Deep Learning Healthcare AI Framework for Real-Time Human Pose Estimation and Remote Patient Monitoring to Support TKR Physiotherapy

Conference paper


AbouGrad, H., Yegamati, M. and Mather, M. 2025. Hybrid Deep Learning Healthcare AI Framework for Real-Time Human Pose Estimation and Remote Patient Monitoring to Support TKR Physiotherapy. 14th International Conference on Human Interaction & Emerging Technologies: Future Systems and Artificial Intelligence Applications. University of East London, London, United Kingdom 10 - 12 Jun 2025 AHFE Open Access.
AuthorsAbouGrad, H., Yegamati, M. and Mather, M.
TypeConference paper
Abstract

Total Knee Replacement (TKR) rehabilitation critically depends on precise physiotherapy exercise execution, and the rise of patient volumes and constrained clinical resources limit continuous supervision. This study presents an Artificial Intelligence (AI) framework for real-time assessment and feedback of TKR exercises using deep learning–based human pose estimation to empower remote rehabilitation. We investigate three architectures: a Dense Convolutional Neural Network (DCNN) incorporating frame decoupling for robust joint tracking; a pruned Generative Adversarial Network (Sparse GAN) optimised for computational efficiency; and a novel hybrid model that embeds the DCNN as a discriminator within the GAN model. A diverse dataset of over 10,000 annotated video clips, sourced from clinical environments and public repositories, was processed with OpenCV, and joint annotations were generated using OpenPose. Models were trained and evaluated on standard metrics (i.e. Precision, Recall, F1-score) alongside runtime and memory usage benchmarks. The hybrid architecture achieved the highest classification performance with 86.01% F1-score, which demonstrates the synergetic benefits of combining rich feature extraction with generative refinement, though it incurred elevated computational costs. The Sparse GAN provided faster inference suitable for mobile deployment, with only a marginal decrease in F1-score. The standalone DCNN provided a balance between accuracy and speed, but it did not match the hybrid’s precision. These results highlight a fundamental trade-off between model complexity and real-time usability in AI-driven therapeutic monitoring. The hybrid model is optimal for clinical settings where accuracy is paramount, while the Sparse GAN provides a practical solution for resource-constrained and edge-based applications. Future work will explore model compression, hardware acceleration, and edge-computing strategies to further optimise performance. By demonstrating the viability of advanced pose estimation techniques in a physiotherapy context, this research contributes to the broader discourse on the use of AI in healthcare for scalable, autonomous rehabilitation tools across several medical and wellbeing domains.

KeywordsHealthcare AI Framework; Remote Rehabilitation; Physiotherapy Patient Monitoring; MedTech; Total Knee Replacement; Dense Convolutional Neural Network; Hybrid DCNN+GAN model
Year2025
Conference14th International Conference on Human Interaction & Emerging Technologies: Future Systems and Artificial Intelligence Applications
PublisherAHFE Open Access
Accepted author manuscript
License
File Access Level
Anyone
Publication process dates
Accepted04 May 2025
Completed12 Jun 2025
Deposited13 May 2025
Journal citationp. In press
Web address (URL)https://www.ihiet-fs.org/program.html
Copyright holder© 2025 The Authors
Permalink -

https://repository.uel.ac.uk/item/8z7x6

Restricted files

Accepted author manuscript

  • 36
    total views
  • 1
    total downloads
  • 9
    views this month
  • 0
    downloads this month

Export as

Related outputs

Financial Decision-Making AI-Framework to Predict Stock Price Using LSTM Algorithm and NLP-Driven Sentiment Analysis Model
AbouGrad, H., Qadoos, A. and Sankuru, L. 2025. Financial Decision-Making AI-Framework to Predict Stock Price Using LSTM Algorithm and NLP-Driven Sentiment Analysis Model.
Efficient Blind Equalizer Schemes Using Variable Tap-length Algorithm
Alsaid, S. and AbouGrad, H. 2025. Efficient Blind Equalizer Schemes Using Variable Tap-length Algorithm. AI and IoT for Next-Generation Smart Robotic Systems Innovations, Challenges, and Opportunities – AISRS Workshop, 3rd International Conference on Mechatronics and Smart Systems – CONF-MSS 2025. University of East London 09 - 09 Dec 2024 EWA Publishing.
Neural Networks Stock Market Price Forecasting Model: Integrating Economic Indicators and Investment Technical Analysis Toward Advanced Financial Analysis
AbouGrad, H., Sankuru, L. and Qadoos, A. 2025. Neural Networks Stock Market Price Forecasting Model: Integrating Economic Indicators and Investment Technical Analysis Toward Advanced Financial Analysis. AI and IoT for Next-Generation Smart Robotic Systems Innovations, Challenges, and Opportunities – AISRS Workshop, 3rd International Conference on Mechatronics and Smart Systems – CONF-MSS 2025. University of East London 09 - 09 Dec 2024 EWA Publishing.
AI-GeoInfo Crop Recommendation Framework Using Decision Tree Classifier and Flask-based GeoAPIs
AbouGrad, H. and Yegamati, M. 2025. AI-GeoInfo Crop Recommendation Framework Using Decision Tree Classifier and Flask-based GeoAPIs. AI and IoT for Next-Generation Smart Robotic Systems Innovations, Challenges, and Opportunities – AISRS Workshop, 3rd International Conference on Mechatronics and Smart Systems – CONF-MSS 2025. University of East London 09 - 09 Dec 2024 EWA Publishing.
AI Investment Advisory: Examining Robo-Advisor Adoption Using Financial Literacy and Investment Experience Variables
Qadoos, A., AbouGrad, H., Wall, J. and Sharif, S. 2025. AI Investment Advisory: Examining Robo-Advisor Adoption Using Financial Literacy and Investment Experience Variables. AI and IoT for Next-Generation Smart Robotic Systems Innovations, Challenges, and Opportunities – AISRS Workshop, 3rd International Conference on Mechatronics and Smart Systems – CONF-MSS 2025. University of East London 09 - 09 Dec 2024 EWA Publishing. https://doi.org/10.54254/2755-2721/2025.21287
AI-Framework to Detect eCommerce Fake Reviews: A Hybrid Neural Network Machine Learning Model
AbouGrad, H. and Shabarshov, A. 2024. AI-Framework to Detect eCommerce Fake Reviews: A Hybrid Neural Network Machine Learning Model. Artificial Intelligence and Computational Technologies: Innovations, Usage Cases, and Ethical Considerations. Near East University (NEU),Turkey 25 - 26 Nov 2024 Springer Nature.
Vehicle-camel collision avoidance system in Libyan desert roads using computer vision technique
Alsaid, S., AbouGrad, H., Abubahia, A. and Ali, N. 2024. Vehicle-camel collision avoidance system in Libyan desert roads using computer vision technique. Sebha University Conference Proceedings. 3 (2), pp. 554-558. https://doi.org/10.51984/sucp.v3i2.3416
Blockchain-based Decentralised Application (DApp) towards Achieving a Hunger-Free World
Auwal, B., Duta, L., Sharif, S., Abou-Grad, H. and Ali, A 2024. Blockchain-based Decentralised Application (DApp) towards Achieving a Hunger-Free World. ICSDI 2024: 2nd International Conference on Sustainability: Development and Innovation. Riyadh, KSA 18 - 22 Feb 2024 Springer. https://doi.org/10.1007/978-981-97-8345-8_54
Decision Making by Applying Machine Learning Techniques to Mitigate Spam SMS Attacks
Abou-Grad, H., Chakhar, S. and Abubahia, A. 2023. Decision Making by Applying Machine Learning Techniques to Mitigate Spam SMS Attacks. 4th International Conference on Deep Learning, Artificial Intelligence and Robotics, (ICDLAIR) 2022. Salerno, Italy 14 - 21 Dec 2022 Springer, Cham. https://doi.org/10.1007/978-3-031-30396-8_14
Applying the Delphi Method to Measure Enterprise Content Management Workflow System Performance
Abou-Grad, H. and Warwick, J. 2022. Applying the Delphi Method to Measure Enterprise Content Management Workflow System Performance. Science and Information Conferences: Computing Conference 2022. 14 - 15 Jul 2022 Springer, Cham. https://doi.org/10.1007/978-3-031-10464-0_27
Developing the Business Process Management Performance of an Information System Using the Delphi Study Technique
Abou-Grad, H., Warwick, J. and Desta, A. 2019. Developing the Business Process Management Performance of an Information System Using the Delphi Study Technique. EAI International Conference on Technology, Innovation, Entrepreneurship and Education (2017). Canterbury, UK 11 - 12 Sep 2017 Springer. https://doi.org/10.1007/978-3-030-02242-6_15