A Method for Locomotion Mode Identification Using Muscle Synergies.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Active lower limb transfemoral prostheses enable diverse locomotion modes.
• Seamless transitions between modes require accurate identification of user intent.
• Current methods often rely on residual limb neural information or prosthesis sensors.

Purpose of the Study:

• To present and evaluate a muscle synergy-based approach for classifying locomotion modes in transfemoral amputees.
• To decode neural information from electromyography (EMG) signals at critical gait instances.
• To compare the performance of the muscle synergy approach against traditional machine learning algorithms.

Main Methods:

• Utilized non-negative matrix factorization (NMF) to extract muscle synergies from EMG data.
• Employed non-negative least squares (NNLS) for neural command estimation.
• Compared the muscle synergy method with linear discriminant analysis (LDA), support vector machine (SVM), and neural network (NN) on able-bodied subjects.

Main Results:

• No significant difference in classification errors between the muscle synergy approach and other methods during the stance phase.
• The muscle synergy approach outperformed NN and LDA during the swing phase.
• Performance of the muscle synergy approach was comparable to SVM during the swing phase.

Conclusions:

• The muscle synergy approach demonstrates potential for discriminating between locomotion modes, including those involving transitions.
• This method offers a viable alternative for enhancing prosthetic control and user experience.
• Further research can explore its application in real-world amputee populations.