CCADD has been selected for the 2026 Internal Research Grant (Track I) by Seoul National University Hospital (SNUH). The project aims to develop COMET-DDI, a novel framework for predicting complex drug-drug interactions (cDDIs), identifying their underlying mechanisms, estimating associated pharmacokinetic/pharmacodynamic (PK/PD) changes, and generating clinically interpretable evidence for each prediction.

Building on the team's prior work in single drug-pair DDI prediction, the project will integrate data from multiple pharmaceutical and biomedical databases to construct a cDDI-specific knowledge graph. Leveraging this foundation, COMET-DDI will employ a multi-agent architecture that combines external biomedical knowledge, specialized AI agents, and a clinician reasoning . By applying recent advances in large language model-based multi-agent systems together with advanced multi-hop reasoning strategies, the framework will jointly model multi-drug combinations and multiple overlapping interaction mechanisms while improving predictive accuracy and mechanistic interpretability.

Prof. Howard Lee and his team anticipate that COMET-DDI will help establish a new paradigm for cDDI prediction. Furthermore, the resulting knowledge graph and benchmark datasets are expected to serve as valuable community resources that support future research in the field.