Predicting Successful Neurofeedback Training using EEG Signals of Primary Sessions

Neurofeedback has an important role in the improvement of cognitive performance in both clinical and healthy individuals. In a neurofeedback process, the person learns how to self-regulate their brain activity and tries to alter their mental state to achieve a desirable brainwave patterns. However, some individuals never learn how to modify their brain activities through neurofeedback training. In this study, we grouped participants as ‘‘performers’’ or ‘‘non-performers’’, based on their ability or inability to modify their brain activity. Then, with feature extraction from early training sessions and using a classifier, performers were classified from non-performers. The results showed that using multilayer perceptron neural network and with two features extracted from EEG signals of fourth neurofeedback session, with 94% accuracy on training data and 82.5% accuracy on test data, performers can be distinguished from non-performers.