Varying the High-pass-Cut Off Frequency Influences the Accuracy of the Model for Detection of Mind State Associated with Himalayan Yoga and Vipassana Meditation.
Annals of neurosciences July 19, 2025 Ritu Munjal, Tarun Varshney
Meditation and yoga practices are increasingly used to prevent ailments. This work examined how the high-pass filter (HPF) cutoff frequency affects single-trial classification accuracy for distinguishing meditative from non-meditative states using electroencephalogram (EEG) data. Two meditation types were studied: Vipassana and Himalayan Yoga. Inception Convolutional Gated Recurrent Neural Network (IC-RNN) and Convolutional Neural Network (CNN) models were compared at various HPF settings. For Vipassana, the highest accuracy was 86.19% with IC-RNN and 99.45% with CNN at a 1 Hz filter. For Himalayan Yoga, the highest accuracy was 88.15% with IC-RNN and 100% with CNN at the same 1 Hz setting. The 1 Hz HPF consistently yielded strong results, suggesting guidelines for filter settings to improve model performance.