An EEG Classifier to Discriminate Between Focused Attention Meditation and Problem-solving
IEEE International Conference on Systems, Man and Cybernetics October 9, 2022 Gansheng Tan, Shui-Bo Wang, Valentin Vierge et al.
A Random Forest classifier trained on two-second samples of EEG data can discriminate between Focused Attention Meditation (FAM) and a problem-solving task with high accuracy when personalized to each individual. Individual classifiers achieved an average accuracy of 93% across 14 subjects, whereas general classifiers trained on inter-individual data performed worse (74% and 54% depending on whether the tested subject's data was included in training). The most discriminating EEG features were Beta mean band amplitude and Theta-Gamma phase-amplitude coupling, especially in occipital and left centro-temporal brain regions. The findings favor personalized classifiers for real-time detection of meditative states.