BLUE: Bi-layer Heterogeneous Graph Fusion Network for Avian Influenza Forecasting

2025-05-28

AI Summary

Accurately predicting avian influenza outbreaks in wild birds requires modeling complex, multi-scale, and cross-regional transmission mechanisms. Existing spatiotemporal graph neural network (GNN) models predominantly rely on geographic adjacency, neglecting critical epidemiological relationships—such as viral genetic lineage tracing—that govern pathogen spread. To address this, we propose a novel two-layer heterogeneous graph modeling framework that explicitly integrates geographic adjacency and genetic溯源 (lineage-based) associations as distinct, non-homogeneous edge types, preserving the structural integrity of each relation. Our approach synergistically combines heterogeneous graph convolution, relation-aware smoothing, and autoregressive graph sequence modeling to enable joint learning from multi-source heterogeneous data—including genomic, spatial, and ecological features. Evaluated on the newly curated Avian-US dataset, our method achieves significant improvements in long-term outbreak prediction accuracy and enhances interpretability of inferred transmission pathways.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks

Abstract

Accurate forecasting of avian influenza outbreaks within wild bird populations requires models that account for complex, multi-scale transmission patterns driven by various factors. Spatio-temporal GNN-based models have recently gained traction for infection forecasting due to their ability to capture relations and flow between spatial regions, but most existing frameworks rely solely on spatial connections and their connections. This overlooks valuable genetic information at the case level, such as cases in one region being genetically descended from strains in another, which is essential for understanding how infectious diseases spread through epidemiological linkages beyond geography. We address this gap with BLUE, a B}i-Layer heterogeneous graph fUsion nEtwork designed to integrate genetic, spatial, and ecological data for accurate outbreak forecasting. The framework 1) builds heterogeneous graphs from multiple information sources and multiple layers, 2) smooths across relation types, 3) performs fusion while retaining structural patterns, and 4) predicts future outbreaks via an autoregressive graph sequence model that captures transmission dynamics over time. To facilitate further research, we introduce extbf{Avian-US} dataset, the dataset for avian influenza outbreak forecasting in the United States, incorporating genetic, spatial, and ecological data across locations. BLUE achieves superior performance over existing baselines, highlighting the value of incorporating multi-layer information into infectious disease forecasting.

Problem

• Modeling multi-scale avian influenza transmission patterns
• Integrating genetic and spatial data for outbreak forecasting
• Capturing epidemiological linkages beyond geographical connections

Innovation

• Bi-layer heterogeneous graph fusion network
• Integrates genetic, spatial, ecological data
• Autoregressive graph sequence model