Genetics Encoded Joint Embedding of Multimodal Connectomes with Explainable Graph Neural Network for Schizophrenia Classification

Investigating structural and functional brain connectivity is crucial for understanding neuropsychiatric disorders like schizophrenia (SZ), where genetic markers such as SNPs (single nucleotide polymorphisms) also play a significant role. Graph Neural Networks (GNNs) offer powerful tools for learning complex patterns from connectome data, yet their use across multimodal connectomes remains under-explored. In this study, we developed an explainable multiview GNN framework for SZ classification, integrating genetic markers with connectome features by leveraging message-passing scheme of GNN with Deep Canonical Correlation Analysis (DCCA) for multimodal fusion. Experiments on clinical dataset confirmed the robustness of the framework, highlighting potential clinical biomarkers.