Classification of gait phases based on a machine learning approach using muscle synergy

Area of Science:

• Biomechanics
• Neuroscience
• Machine Learning

Background:

• Accurate gait phase detection is vital for diagnosing neurological/musculoskeletal disorders and controlling assistive devices.
• Muscle co-activation patterns, known as muscle synergies, are useful for understanding locomotion and rehabilitation.

Purpose of the Study:

• To investigate if muscle synergy features provide more accurate and robust gait event classification compared to traditional time-domain and wavelet features.
• To analyze the functional organization of the human movement system during gait.

Main Methods:

• Electromyography (EMG) signals were collected from eight healthy individuals walking.
• Non-negative matrix factorization (NNMF) identified muscle synergies.
• Gait phases were classified using Decision Tree, KNN, SVM, and Neural Network algorithms.

Main Results:

• Muscle synergy features demonstrated superior classification accuracy for gait phases compared to traditional EMG features.
• The findings support the hypothesis that muscle synergies enable precise gait phase classification.

Conclusions:

• Muscle synergy analysis presents a novel and effective approach for gait analysis.
• Muscle synergies show significant potential as a tool for accurate gait phase detection in clinical and assistive device applications.