Efficient, Accurate and Interpretable Drug-Drug Interaction Prediction Using Group Graph-Based Models
Year: 2026
Type: conference-paper
Abstract: Drug-drug interaction (DDI) prediction is crucial for drug safety, yet traditional methods often face challenges in interpretability and accuracy. Meaningful molecular substructures are key to improving both the performance and explainability of DDI prediction. To this end, we propose the use of a group graph who is a molecular graph in which nodes represent chemically meaningful substructures such as functional groups to address these limitations. Experimental results on benchmark datasets, show that models based on group graph, achieve predictive performance comparable to or slightly better than counterparts based on atom graph. Moreover, models based on group graph improves computational efficiency by 20% to 50%. Importantly, models based on the group graph offer enhanced interpretability: the identified critical substructures correspond to complete functional groups (e.g., aromatic rings), which are chemically intuitive and consistent with known pharmacological mechanisms.
Language: English
Subfield: Drug Discovery
Domain: Life Sciences
Open Access status: hybrid
OpenAlex ID: W7203976610