Classification of mindfulness experiences from gamma-band effective connectivity: Application of machine-learning algorithms on resting, breathing, and body scan.
Computer methods and programs in biomedicine December 1, 2024 Ai-Ling Hsu, Chun-Yu Wu, Hei-Yin Hydra Ng et al.
Electroencephalography (EEG) effective connectivity can predict whether someone has experience with mindfulness-based stress reduction (MBSR). Machine learning algorithms classified participants' MBSR history using gamma-band brain connectivity. The decision tree algorithm achieved the highest prediction accuracy of 91.7% during resting state, outperforming classifications during focus-breathing and body-scan sessions. Support vector machine and naïve Bayes classifiers also showed significant accuracies above chance across all three sessions. Preserving just four EEG channels (F7, F8, T7, P7) out of 19 yielded 83.3% accuracy. Connectivity features predominantly in the frontal lobe contributed most to classifier construction, consistent with existing mindfulness literature.