MARD: Mirror-Augmented Reasoning Distillation for Mechanism-Level Drug-Drug Interaction Prediction

In three linesMARD is a 7B parameter model for mechanism-level drug-drug interaction prediction (enzyme, pharmacodynamic axis). Uses reasoning distillation with process-reward-weighted DPO and mechanism-aware retrieval. On April-2026 DrugBank: +13.9pp over best baseline, +6.7pp over GPT-4o, with robust generalization to unseen drug pairs.

Summary generated by Claude — human-verified