Novel Framework for Stroke Rehabilitation Using High-Density Surface Electromyography Signals and Improved Whale Optimization Algorithm-based LSTM Network

Strategic Insights -

Researchers from Guizhou Normal University have developed a novel framework for stroke rehabilitation, utilizing high-density surface electromyography (HD-sEMG) signals and an improved whale optimization algorithm (IWOA)-optimized long short-term memory (LSTM) network. The framework aims to address the limitation of traditional sEMG analysis, which often overlooks the complexity of object-interactive behaviors crucial for patient independence. By integrating HD-sEMG signals with an IWOA-optimized LSTM network, the researchers have achieved superior performance in self-care behavior recognition and continuous gesture recognition, with 99.58% and 86.19% accuracy, respectively.

Key Takeaways:

  • The proposed framework combines HD-sEMG signals with an IWOA-optimized LSTM network to address the limitation of traditional sEMG analysis.
  • The framework offers a robust, scalable solution for clinical applications, bridging the gap between laboratory-based gesture recognition and practical, patient-centered care.
  • Experimental results demonstrate superior performance, achieving 99.58% accuracy in self-care behavior recognition and 86.19% accuracy for 17 continuous gestures on the Ninapro db2 benchmark.
  • The framework operates with low latency, meeting the real-time requirements for assistive devices.
  • The proposed methodology enables precise, context-aware recognition of daily activities, advancing personalized rehabilitation technologies and empowering stroke patients to regain autonomy in self-care tasks.
  • The researchers created a specialized HD-sEMG dataset capturing nine continuous self-care behaviors, along with time and posture markers, to better reflect real-world patient interactions.
  • The framework's multi-channel feature fusion module based on Pascal's theorem enables efficient signal segmentation and spatial-temporal feature extraction.
  • The researchers enhanced the IWOA algorithm, integrating optimal point set initialization, a diversity-driven pooling mechanism, and cosine-based differential evolution to optimize LSTM hyperparameters, improving convergence and global search capabilities.

Statistics:

  • 99.58% accuracy in self-care behavior recognition
  • 86.19% accuracy for 17 continuous gestures on the Ninapro db2 benchmark
  • Low latency of the framework, meeting real-time requirements for assistive devices
  • 9 continuous self-care behaviors captured in the HD-sEMG dataset
  • 17 continuous gestures recognized by the framework
  • 152 Beach Road, #21-01 Gateway East, Singapore, Singapore - Location of Springer Singapore Pte Ltd

Sources:

  • NewsRx LLC. Researchers from Guizhou Normal University Describe Findings in Genomics and Genetics (Prediction of Self-care Behaviors In Patients Using High-density Surface Electromyography Signals and an Improved Whale Optimization Algorithm-based Lstm ...). Health & Medicine Week. June 20, 2025; p 4391.
  • Journal of Bionic Engineering. Prediction of Self-care Behaviors In Patients Using High-density Surface Electromyography Signals and an Improved Whale Optimization Algorithm-based Lstm Model. 2025.
  • Springer Singapore Pte Ltd. Journal of Bionic Engineering. 152 Beach Road, #21-01 Gateway East, Singapore, Singapore.
  • Elsevier. Journal of Bionic Engineering. www.journals.elsevier.com/journal-of-bionic-engineering/.