A Machine Learning based Approach for Classification of a Person's Actions based on Electromyography (EMG) Signals

The purpose of this research is to utilize a QuestNET (QNET) Myoelectric Board in conjunction with an (National Instruments) (NI) Elvis II+ board to assess the electrical activity of a user's muscle tissue. LabVIEW, which stands for Laboratory Virtual Instrument Engineering Workbench, is used to connect the board to the computer. LabVIEW is a graphical language that allows the user and the board to interact directly. The data is gathered from the primary source, 5 statistical measures are implemented on the data, graphs are plotted according to the time stamp for a male subject (age range: 20–25), and the studied data is summarized to make important conclusions. A dataset was then created using the windowing technique to train the classification model and prediction results were analyzed by implementing different prediction algorithms like K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Decision Trees. The accuracy of predicting the actions of the user based on the electrical activity of the muscle tissues has been formulated as results and necessary conclusions are drawn to show that KNN algorithm bears the highest accuracy in prediction with an accuracy of 95%.