Analysis of Neural Network based Proportional Myoelectric Hand Prosthesis Control

We show that state-of-the-art advanced artificial intelligence algorithms based on “deep neural networks” achieve superior performance compared to two common algorithms for decoding movement intent for prosthetic control. We performed a rigorous analysis to explain why the state-of-the-art algorithms perform better based on an analysis of 11 able-bodied subjects and 4 prosthesis users. This work shows and explains why state-of-the-art algorithms are better which is important since it has been a challenge to demonstrate that new algorithms are better due to external disturbances.