High-Pass-Filtering Impacts the Accuracy of the Hybrid Model for Detection of Brain State Allied with Isha Shoonya Meditation

Background: Many psychological and physical health problems are instigated due to existence of stress because of chaotic lifestyle. It is a noteworthy hazardous factor for several cerebral disorders and hence, impacts the superiority of life. To manage the stress, it is important to practice the meditation and yoga. Electroencephalogram (EEG) is the chosen analytical tool in this research as it is economical, non-invasive and easy to operate. This study mainly focuses on Deep Learning (DL) technique for recognition of brain state linked with Isha Shoonya meditation. To obtain the best outcomes, the task was accomplished for identifying meditative brain state with varying High Pass Filter (HPF) frequencies.

Purpose: The performance of the model depends upon various factors like model designing, hyperparameter tuning, Independent Components Analysis (ICA), and HPF. This study emphasized on one important factor how the high pass filtering impacts the accuracy of the model. The filter setting was done to dissimilar frequencies: 0.1 Hz, 0.5 Hz, 1 Hz, and 2 Hz to investigate the varying impacts of HPF on the presented Hybrid model. The performance was systematically assessed by varying the filter settings.

Methods: Hybrid model was designed and examined how the high pass filtering impacts the accuracy of the model. Results and

Conclusion: Accuracy of 68.93% with filter setting at 0.1 Hz, 87.50% at 0.5 Hz, 96.41% at 1 Hz, and 93.76% at 2 Hz was attained. The maximum accuracy of 96.41% was achieved for Hybrid model at 1 Hz for Isha Shoonya meditation. HPF at 1 Hz gave decent outcomes.