Effect of User Practice on Prosthetic Finger Control With an Intuitive Myoelectric Decoder

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Robotics

Background:

• Machine learning-based myoelectric control aims to mimic natural arm movement by mapping muscle signals to prosthesis control.
• User adaptation and performance improvement have been observed in classification-based myoelectric control systems.
• The impact of user adaptation on continuous, dexterous myoelectric prosthesis control remains largely unexplored.

Purpose of the Study:

• To investigate the effect of short-term user adaptation on myoelectric control performance in dexterous tasks.
• To assess whether intuitive decoders are sufficient to overcome control challenges through experience.
• To evaluate the reliability of offline analyses in predicting real-time myoelectric control performance.

Main Methods:

• Real-time experiments involving independent finger position control were conducted.
• Ten able-bodied and two transradial amputee subjects participated in the study.
• Performance metrics were analyzed to quantify improvements due to user experience.

Main Results:

• Significant performance improvements were observed with experience, despite the use of an intuitive decoder.
• Challenges in achieving natural control were attributed to anatomical differences, decoding inaccuracies, and lack of proprioception.
• Offline analysis methods were found to be unreliable predictors of real-time performance.

Conclusions:

• User adaptation plays a crucial role in enhancing myoelectric control performance, even in intuitive systems.
• Real-time experimental validation is essential for accurately assessing the efficacy of myoelectric control algorithms.
• Further research is needed to address inherent control challenges for more natural prosthesis integration.