Methods of Processing and Analyzing EEG Data Based on Machine Learning to Determine Phobic Disorders

The study considers methods for processing and analyzing medical data for the objective diagnosis of phobic disorders using virtual reality. Resulting in the development of an approach to the analysis of electroencephalogram data using machine learning. For which the collection and labeling of EEG data from participants in the control group during the exercise on fear of heights in virtual reality was necessary. The task of choosing the optimal algorithm for classifying data for the presence of a phobic reaction is set. Several architectures of neural networks were compared, and convolutional neural networks trained on an extended amount of data showed the best results (accuracy up to 87%). The developed software can be used to automate the detection of a phobic reaction using virtual reality in the development of new clinical methods for diagnosing phobic disorders.