CLASSIFYING EEG SIGNAL SEGMENTS USING MACHINE LEARNING

In this paper we approach the classification of electroencephalogram (EEG) signals segments with the purpose of predicting movement. Labeling these segments is of utmost importance for patients that have motor functions impairments that affect their lifestyle. We aim to create the artificial intelligence (AI) base for an application that could help patients move a cursor on screen using their brain activity captured by a series on electrodes placed on the surface of their scalp. Machine learning (ML), even though largely considered a simple way to solve such problems, proved to be largely accurate. The main idea of the present work is using the electric potential difference between the brain's hemispheres, as both global maxima and P300 features extracted from C3 and C4 channels, fed to a Random Forest Classifier, as a way to predict movement intentions.