Investigation of the best feature selection in the machine learning based classification of electroencephalography signs

Anxiety affects productivity, quality of life, human abilities and behaviors. It can be considered the main cause of depression and suicide. Clinicians today use specific criteria to diagnose anxiety disorders. There is a need for reliable, non-invasive techniques to perform the complex anxiety detection task. This study aimed to classify binary and four-class categories by analyzing electroencephalography (EEG) signals with fewer EEG channels and features. The DASPS database was used, containing EEG signals from 23 individuals with 14 channels. Using EEGLAB, 4 channels were selected from 14 channels. The wavelet transform, Fourier transform, and statistics were used for feature extraction. The classification was performed using four methods from the MATLAB Classification Learner Toolbox. The highest accuracy rates were achieved with support vector machines for binary classification with 68.3% accuracy and with support vector machines for four-class classification with 61% accuracy.