federated-brain-trajectory-gnn

What problem does it solve?

Federated multi-trajectory GNNs enable privacy-preserving prediction of infant brain connectivity evolution across multiple centers.

Core Features & Use Cases

• Multi-modal trajectory modeling: handles T1-w, T2-w, DTI brain connectivity data.
• Federated training: data never leaves hospitals; aggregates model parameters.
• Two-step imputation + auxiliary regularization: improves performance with sparse data and longitudinal consistency.
• Use Case: multi-center infant brain development studies with limited samples.

Quick Start

Initiate a federated training across multiple hospitals to forecast future infant brain connectivity trajectories with FedGmTE_Net.