Improving myoelectric pattern recognition positional robustness using advanced training protocols.

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Neuroprosthetics

Background:

• Surface electromyogram (EMG) control is crucial for powered upper limb prostheses.
• Current prosthetic hands lack control schemes to fully utilize their mechanical dexterity.
• Pattern recognition for EMG-based control faces challenges with varying prosthesis positions during daily activities.

Purpose of the Study:

• To investigate the impact of varying prosthesis positions on pattern recognition control.
• To evaluate the effectiveness of training strategies for improving positional robustness.
• To determine if dynamic activity training enhances control performance without extended training periods.

Main Methods:

• Assessed pattern recognition performance degradation across static and dynamic positioning tasks.
• Compared control accuracy using prostheses in different spatial orientations.
• Investigated the effect of training with dynamic activities versus static training.

Main Results:

• Pattern recognition control accuracy degrades significantly with varying prosthesis positions.
• Training with dynamic activities demonstrated improved positional robustness for both static and dynamic tasks.
• Enhanced robustness was achieved without necessitating complex or lengthy training protocols.

Conclusions:

• Dynamic activity-based training is a viable strategy to improve the positional robustness of EMG pattern recognition for upper limb prostheses.
• This approach enhances the practical usability of advanced prosthetic hands in activities of daily living.
• Future research should focus on optimizing dynamic training protocols for diverse user needs.