State of the Art Methods of Machine Learning for Prosthetic Hand Development: A Review

Abstract

In developing countries, a number of upper limb incidence that leads to trauma and amputation is increasing. The development of prosthetic hands has been carried out by several researchers in the past, but research related to prosthetic hands that are light, low battery consumption and high accuracy are still a challenge for researchers. Therefore, this study aims to review papers related to the EMG pattern recognition, analog part, feature extraction and classifier methods to get the best prosthetic hand design recommendations. This review paper collects articles from the Scopus and PUBMED databases from 2012 to 2022. The keywords used are EMG AND pattern recognition AND prosthetic hand. Based on the analysis of the VOSviewer application with these keywords, this topic is grouped into five network clusters. Based on the literature study related to the embedded system platforms used, it was found that 27.7% used microcontroller platforms, 11.11% FPGA platforms, 27.7% Raspberry Pi platforms and 33.5% used computer platforms. Furthermore, conventional supervised machine learning is more widely used as a classifier including decision trees, random forests, K-NN, and support vector machines than deep learning. This review paper can provide an overview of the state-of-the-art methods used in the development of machine learning-based smart prosthetic hands.