Interpreting and Analyzing Electric Signals for Brain-Computer Interface of a Brain-Controlled Wheelchair using Machine Learning

In this study, a system that extracts, reads, and analyzes brain signals to convert them into usable commands, particularly for a brain-controlled wheelchair, was successfully demonstrated using machine learning. Electroencephalography (EEG) is a technique that allows scientists and engineers to read electrical signals caused by brain activity. Advancements in EEG technology allow for its implementation such as brain-computer interfaces (BCI), wherein the electronic signals gathered from the EEG are processed as commands to perform certain outputs. Analysis of electrical signals is mainly done through the use of artificial intelligence, where the results of different methods are compared to isolate the best method. These methods include the FastAI neural network, XGBoost, and the random forest model. From the results, the FastAI neural network was the most accurate, with 82% accuracy for simulated testing and an accuracy of 81% for real-time testing. As such, a system that utilizes FastAI neural network is a viable option in translating EEG signals into directional commands.