Neural Signal Decoding for Prosthetic Control: Brain- and Muscle-Signal-Based Approaches

Abstract

Neural decoding technology has made great strides in the past two decades, transitioning from the lab to clinical applications. Brain signals and electromyographic (EMG) signals have emerged as two key information sources for prosthetic control: the former includes invasive and noninvasive neural electrical activity, and the latter leverages residual muscle and peripheral nerve signals. This paper reviews recent advances in prosthetic control from the perspectives of brain-signal-based control and musclesignal-based control. This paper introduces the main signal types, decoding algorithms, and representative applications, provides a comparative analysis of these signals in terms of information capacity, stability, invasiveness, etc., and discusses current limitations and future prospects. By surveying the latest research, this review aims to offer valuable insights for future neuroprosthetics research and clinical applications. At a broader level, the development of neural decoding reflects the ongoing convergence of science, technology, and medicine, and its future progress will not only reshape prosthetic design but also expand our understanding of human–machine interaction, rehabilitation, and even the nature of human agency itself.