Reproduce “Causal knowledge graph analysis identifies adverse drug effects” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sumyyah Toonsi, Paul Schofield, Robert Hoehndorf
Researchers at King Abdullah University of Science and Technology developed Causal Knowledge Graphs (CKGs), a framework that formally integrates knowledge graphs with structural causal models to enable knowledge-driven causal inference. Applying this to pharmacovigilance, the CKG methodology successfully identifies adverse drug effects, improving the prediction of shared drug indications by 1.9% ROC AUC when combined with existing databases.
Key insights
- CKGs bridge symbolic knowledge representation with statistical causal inference, allowing for automated, knowledge-aligned hypothesis generation and confounder identification.
- The disjunctive cause criterion, combined with LASSO regression for covariate selection, effectively identifies adjustment sets for robust causal mediation analysis in large biomedical datasets.
- Formal integration ensures that probabilistic interpretations and causal inferences respect the semantic and hierarchical relationships encoded within the knowledge graph.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.