The Role of Frontal Electrodes in Human State Anxiety Detection Using Wavelet Features and Machine Learning

Anxiety disorder, an intricate emotional state characterized by intense fear or worry, necessitates appropriate intervention for the promotion of societal well-being. Electroencephalography (EEG) is widely used in neural changes research to identify and categorize anxiety states. In this study, EEG signals obtained from 23 individuals, sourced from the publicly accessible “A Database for Anxious States Based on Psychological Stimulation (DASPS),” were utilized to classify anxiety into two levels. We separated the electrode set into three regions: frontal, parietal, and occipital. From each EEG band, a distinct set of statistical features was extracted. The k- Nearest Neighbor (KNN) was used in the classification with a trial frame of 30 seconds. First, we used EEG signals from all 14 channels to detect anxiety and achieved an accuracy of 79.85%. Second, we performed classification using a database from three regions and obtained accuracy of 84.23%, 78.94%, and 63.15% for the frontal, parietal, and occipital electrodes, respectively. The results provided demonstrate the usefulness of using the front lobe of the brain to detect anxiety disorder, whether in terms of accuracy of the results or the low-cost of the EEG headset.