Treffer: Myoelectric Prosthesis Control using Recurrent Convolutional Neural Network Regression Mitigates the Limb Position Effect.

A recent study published in Medical Devices & Surgical Technology Week discusses the development and testing of two recurrent convolutional neural network (RCNN) controllers for myoelectric prostheses. These controllers aim to address the limitations of current pattern recognition-based controllers, which struggle with control deficiencies in high limb positions and only allow for one movement at a time. The RCNN controllers showed improved functional task performance compared to a commonly used linear discriminant analysis classification controller in simulated prosthesis tasks. This research contributes to the development of accurate and simultaneous control of prostheses using EMG-based technologies. [Extracted from the article]