AUTHOR=Lu Zihe , Wang Jialin , Wang Fengqin , Wu Zhoumin 

TITLE=Application of graph frequency attention convolutional neural networks in depression treatment response

JOURNAL=Frontiers in Psychiatry

VOLUME=Volume 14 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2023.1244208

DOI=10.3389/fpsyt.2023.1244208

ISSN=1664-0640

ABSTRACT=Depression, a globally prevalent mental health disorder, necessitates precise prediction of treatment response to enhance personalized care and patient outcomes. Graph Convolutional Neural Networks (GCNs) have emerged as a promising approach for handling complex signals and classification tasks owing to their end-to-end neural architecture and nonlinear processing capabilities. This study investigates the utilization of graph neural networks in EEG data classification to assess responses to antidepressant treatment, distinguishing between treatmentresistant and treatment-responsive cases. EEG signals are transformed into graphs, depicting connections between electrodes and brain regions. Our model, named the Graph Frequency Attention Convolutional Neural Network (GFACNN), integrates a frequency attention module to highlight brain rhythm information. Experiments conducted on an EEG dataset from Peking University People's Hospital demonstrate that GFACNN excels in discriminating treatment responses in depression patients, outperforming deep learning methods. This underscores the efficacy of graph neural networks in exploiting EEG signal connections. In summary, GFACNN holds promise for classifying depression EEG signals, potentially contributing to clinical diagnosis and treatment.