User training for pattern recognition-based myoelectric prostheses: improving phantom limb movement consistency and...

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Neuroscience

Background:

• Pattern recognition-based myoelectric prostheses offer advanced control for amputees.
• User training is crucial for optimizing the performance of these complex devices.
• Limited longitudinal studies exist on the impact of user training on myoelectric prosthesis control.

Purpose of the Study:

• To assess the impact of a two-week training program on the control of a virtual myoelectric prosthesis by transradial amputees.
• To quantify improvements in prosthesis control measures following user training.
• To investigate the long-term retention of learned control strategies.

Main Methods:

• Four transradial amputees participated in eight one-on-one training sessions using a virtual prosthesis with visual biofeedback.
• Control was assessed based on movement completion percentage, normalized movement completion time, and classifier accuracy.
• High-dimensional data clusters were analyzed to understand individual learning strategies.

Main Results:

• Significant improvements were observed across all control measures: movement completion increased from 70.8% to 99.0%, normalized movement completion time improved from 1.47 to 1.13, and classifier accuracy rose from 77.5% to 94.4% (p<0.001).
• Performance remained largely stable eight hours after training cessation, indicating robust learning.
• Individualized strategies for phantom limb movement control were identified for each participant.

Conclusions:

• User training is highly effective in enhancing the control of virtual myoelectric prostheses for transradial amputees.
• The study demonstrates the potential for significant, lasting improvements in prosthesis functionality through targeted training.
• This research establishes a foundation for future longitudinal studies on user-centered training protocols for advanced prosthetic devices.