Modern biomedical data is inherently interconnected, genes interact with proteins, diseases share pathways and patients relate through phenotypic and molecular similarity. Traditional deep learning models often fail to capture this complex relational structure. This session explores how Graph Neural Networks (GNNs) are transforming biomedical discovery by enabling us to model these intricate connections directly.

The talk will walk through real-world examples of how graph learning can be applied to biological networks and patient stratification. It will also compare GNN-based approaches to conventional deep learning, highlighting where graph-based reasoning provides unique advantages in interpretability and data efficiency. Attendees will gain practical insights into structuring biomedical data as graphs, selecting the right GNN architectures, and evaluating model trustworthiness. By the end, participants will understand how GNNs can help uncover hidden biological relationships that traditional methods often overlook.