Detecting muscle fatigue during lower limb isometric contractions tasks: a machine learning approach

Area of Science:

• Biomedical Engineering
• Sports Science
• Rehabilitation Technology

Background:

• Muscle fatigue is a key indicator of exercise-induced fatigue.
• Electromyography (EMG) is vital for monitoring muscle activity and assessing fatigue.
• Automated detection of muscle fatigue is crucial for performance and safety.

Purpose of the Study:

• To develop and validate a machine learning approach for automatic muscle fatigue detection using EMG signals.
• To enhance the accuracy and efficiency of muscle fatigue assessment.
• To provide a tool for monitoring fatigue in various settings.

Main Methods:

• EMG signals were collected from lower limb muscles during isometric contractions.
• Intrinsic mode functions (IMFs) were extracted using improved complementary ensemble empirical mode decomposition adaptive noise (ICEEMDAN).
• A machine learning model, combining support vector machines (SVM) with ICEEMDAN, was employed for fatigue classification after dimensionality reduction with t-SNE.

Main Results:

• EMG signal characteristics significantly correlated with increasing muscle fatigue.
• The proposed SVM and ICEEMDAN model achieved a high classification accuracy of 99.8%.
• The method demonstrated superior performance compared to current state-of-the-art fatigue detection techniques.

Conclusions:

• The developed machine learning strategy is highly effective for automatic muscle fatigue detection.
• This approach offers a valid and reliable method for monitoring fatigue in training, rehabilitation, and occupational environments.
• The high accuracy suggests significant potential for practical applications in sports science and occupational health.