BRIDGE: Biological Antimicrobial Resistance Inference via Domain-Knowledge Graph Embeddings

BRIDGE predicts antimicrobial resistance links in Klebsiella pneumoniae and Escherichia coli with classification accuracy up to 97% using knowledge graph embeddings.

• Why it matters: AMR surveillance is hindered by labor-intensive ARG detection and limited ability to capture novel resistance mechanisms.
• What they did: The framework builds a biologically grounded knowledge graph from CARD, STRING, and DrugBank, then uses knowledge graph embedding models combined with neural networks, integrating gene neighbourhood and PPI networks.
• The result: Adding gene neighbourhood information and PPIs improves both predictive accuracy and interpretability of AMR link predictions, supporting scalable, data-integrated surveillance and drug discovery.

AI summary of the abstract · may contain errors · not medical advice · read the paper.