Assisting Schizophrenia Diagnosis Using Clinical Electroencephalography And Interpretable Graph Neural Networks: A Real-world And Cross-site Study

This article discusses the development of an advanced interpretable deep learning model using multimodal clinical electroencephalogram (EEG) features and demographic information as inputs to graph neural networks. The model was trained on a small clinical dataset and enhanced using a large-scale public dataset of adult participants. Cross-site validation from independent datasets of adult and adolescent participants produced stable performance, with AUCs of 0.793-0.852 and accuracies of 0.786-0.858. Feature visualization revealed that EEG theta and alpha band power appeared to be the most significant and translational biomarkers of schizophrenia pathology.