Machine Learning Modelling of EEG Bio Signals Using Motor Imagery

This study investigates the potential of Brain Computer Interfaces (BCI) as a novel approach to motor rehabilitation for individuals with severe paralysis resulting from strokes. Traditional rehabilitation methods often fall short in such cases, necessitating innovative solutions. The research utilizes Electroencephalogram (EEG) bio signals during motor imagery tasks to evaluate its effectiveness. This study utilized machine learning algorithms to extract relevant features and classify outcomes. Notably, Random Forest (RF) stands out as a high-performing classifier, achieving an impressive accuracy of 95.79%. The strategic combination of AdaBoost M1 with RF further enhances performance, yielding a slightly higher accuracy of 95.87% with reduced execution time (25.58 seconds). These findings suggest the potential of BCI technologies, particularly when coupled with machine learning, to offer targeted and efficient rehabilitation interventions for individuals with limited or no motor function after a stroke.