Covariate shift adaptation in EMG pattern recognition for prosthetic device control.

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Signal Processing

Background:

• Myocontrol algorithms for prosthetics face robustness challenges due to nonstationarities like electrode shifts and varying physiological conditions.
• These signal variations, termed covariate shift, degrade classification accuracy in real-world applications.
• Traditional retraining methods are time-consuming, requiring extensive data collection.

Purpose of the Study:

• To develop and evaluate a novel method for adapting electromyography (EMG) classifiers using minimal calibration data.
• To enhance the robustness of myocontrol systems against nonstationary signal variations.
• To reduce the time and data required for retraining EMG classifiers.

Main Methods:

• Proposed an adaptation strategy for EMG classifiers using a small calibration set to capture nonstationarities.
• Implemented an estimator that shrinks training model parameters towards calibration set parameters.
• Tested the strategy on EMG signals acquired over 5 days from able-bodied individuals.

Main Results:

• The proposed shrinkage estimator significantly improved classifier performance across different testing days.
• Classification accuracy remained above 92% over five days, even with only one trial per class for retraining.
• The method effectively captured relevant nonstationary aspects of the EMG signals.

Conclusions:

• The proposed methodology offers a practical solution for enhancing the robustness of myocontrol pattern recognition.
• Adapting EMG classifiers with small calibration sets is efficient and effective for daily use.
• This approach can lead to more reliable and user-friendly prosthetic devices.