Exploring Stress Reduction Using EEG Signal Analysis and Machine Learning: Effects of Music and ASMR

Auditory-based stimulation has recently gained attention as a non-invasive approach for stress regulation. This paper presents an EEG-based analysis of stress modulation induced by Autonomous Sensory Meridian Response (ASMR) stimuli and music. EEG signals were recorded from nine subjects during exposure to ASMR-triggering sounds, traditional Iranian music, and classical music. Stress-related neural features were extracted using spectral power analysis in theta, alpha, and beta frequency bands, along with multiscale entropy. Multiple machine learning and deep learning models, including SVM, MLP, CNN, and LSTM, were employed to classify baseline and relaxation states. Experimental results indicate a significant reduction in theta-band power during ASMR and traditional music exposure (p < ۰.۰۵), comparable to classical music. A hybrid CNN–LSTM model achieved the highest classification accuracy of ۹۲.۴%. Time-dependent analysis further showed that maximum stress reduction occurred after approximately ۱۰ minutes of auditory stimulation, highlighting the potential of ASMR-based approaches for EEG-driven stress management systems.

کلیدواژه ها:

EEG Signal Processing ، Machine Learning ، Deep Learning ، Stress Classification ، Auditory Stimuli