Towards the development of personalized and generalized interfaces for brain signals across different styles of meditation
Indian Conference on Computer Vision, Graphics & Image Processing December 8, 2022 Shruti Singh, Pankaja Pandey, Shivam Chaudhary et al.
Machine learning models can distinguish between the neural oscillations of expert meditators and non-meditators using EEG data. The study analyzed EEG recordings from expert practitioners of Himalayan Yoga, Vipassana, and Isha Shoonya, along with non-expert controls. Thirteen different machine learning models were applied for within-subject and cross-subject classification across six conditions for both meditation and mind-wandering. Features extracted from the mean of 64 EEG time series were used. Within-subject classification achieved maximum accuracy. In cross-subject analysis, accuracy reached 18.3% above chance level in meditation between controls and Isha Shoonya, and over 18% above chance in mind-wandering between controls and Vipassana. These results suggest that both personalized and generalized models could guide novice practitioners to modulate brain signals toward expert levels.