Classification of schizophrenia patients using a graph convolutional network: A combined functional MRI and connectomics analysis

•Graph convolutional network was used to automatically diagnose schizophrenia based on brain region and connectivity features.•A quantification method was used to construct sparse functional brain networks.•Implemented generalizability evaluation based on 200 train/test data splits.•Better performance than the traditional machine learning, MLP and CNN methods.•Functional characteristics (ALFF, ReHo, and DC) performed better than node property characterisitics (NE, BC, and DC).