Multi-Omics Risk Stratification in Cancer Genomics
Our Graph Attention Network (GAT/GATv2) integrates multi-omics data with biological knowledge to predict patient survival outcomes and stratify risk groups with unprecedented accuracy.
On selected biomarkers and stratification tasks
Integrates genomics, proteomics, and clinical factors
Highlights influential pathways and interactions
Graphs constructed from known gene/protein interactions and pathways.
Attention over nodes captures cross-modal signal and important subnetworks.
Custom heads for risk scores and clinical endpoint prediction.
