Anxiety Anxiety Detection Based on EEG Signal using 1-D Convolutional Neural Network Classifier

Abstract

Anxiety is defined as fear and symptoms of somatic tension experienced when a threat or danger is anticipated. In recent years, biological markers have been explored to detect anxiety noninvasively, one method is Electroencephalography (EEG). Detecting state anxiety using EEG is an intriguing area of research. This study detects the state of anxiety based on an EEG signal using a 1-Dimensional Convolutional Neural Network (1-D CNN). The dataset is provided by the Database for Anxious States based on Psychological Stimulation (DASPS). DASPS is an EEG recording obtained from twenty-three participants for this investigation. The data were analyzed for statistical features, and then a 1-D CNN was employed to classify anxiety levels. The results show that 95.1% of mild and severe anxious conditions can be accurately detected. Furthermore, 94.8% of detection accuracy is achieved when anxiety is classified as normal, mild, moderate, or severe. Overall, this study provides a solid foundation for multi-level anxiety detection by improving the accuracy and selecting better features.