Automating Biomedical Knowledge Graph Construction For Context-Aware Scientific Inference

Biomedical interactions are inherently dynamic, often shifting or even reversing under specific physiological states. However, existing extraction methods simplify these complex mechanisms into context-agnostic binary associations, resulting in semantic loss and contradictory evidence. Here, we present AutoBioKG, an end-to-end framework that constructs context-aware knowledge graphs by leveraging composite triplets to encode environmental conditions and entity attributes alongside core relationships. Powered by a self-evolving open information extraction model trained solely on our curated BioOpenIE dataset, the framework refines its generalization capabilities. Notably, AutoBioKG surpasses state-of-the-art large language models in zero-shot settings, achieving F1 score improvements ranging from 18.5% to 20.7% across three benchmarks (DDI, ChemProt, and BioRED). Furthermore, AutoBioKG-derived graphs significantly outperform existing approaches in the BioASQ biomedical question-answering task, particularly for complex queries requiring fine-grained contextual information. AutoBioKG offers a scalable and accessible solution for transforming unstructured literature into actionable biomedical knowledge.