Social anxiety disorder (SAD) is a serious concern for medical practitioners worldwide. Identifying its severity level (severe, moderate, mild, or none) can be challenging, which is why this paper proposes a solution.
The researchers developed a method that classifies SAD severity levels using the patterns of brain information flow and graphical network structures, analyzed via an EEG in 66 patients with different SAD severities along with 22 demographically matched healthy controls.
The study examined differences between SAD groups and HCs in varying frequency bands, finding that severe and moderate SAD groups had greater information flow than mild and HC groups in all bands. The study also used machine learning classifiers to distinguish three classes of SAD from HCs. Combining graph theory with PDC values produced optimal results, accurately identifying SAD with an impressive rate of 92.78% and other performance measures.
The findings could lead to new biomarkers for SAD diagnosis using topological brain networks and machine learning. By quantifying directed information flow and utilizing graph theory measures, the researchers are optimistic their findings will help practitioners customize better patient treatment programs.