Bayer: Knowledge graph reasoning and GraphRAG for early target discovery

This video presents two knowledge graph-driven methods—graph reasoning and GraphRAG—for improving early drug target discovery by reducing AI hallucinations and enhancing the accuracy and grounding of target hypotheses.

What it covers

Graph Reasoning
A method for inferring novel gene-disease associations using logical deductions within a knowledge graph.
GraphRAG
A technique that anchors retrieved facts to verifiable graph relations to improve reliability in AI-generated hypotheses.
Reducing Hallucination
Both methods help minimize false or unsupported claims in AI-driven target discovery.
Enhancing Predictive Power
The integration of knowledge graphs improves the accuracy of predicting viable drug targets.
Grounding in Verifiable Relations
Ensures that AI-generated insights are tied to actual, traceable relationships in the knowledge graph.
Early Target Discovery
The approaches are designed to support the generation of testable hypotheses in the early stages of drug development.