Federal Grant Award Summary

Award Details: The Trustees of Boston University (Boston University Medical Campus) received a $449,625 Project Grant from the National Center for Advancing Translational Sciences (NCATS, CFDA 93.350) effective May 1, 2026, through April 30, 2028. The award supports development of AI-driven biomedical knowledge hypergraphs designed to improve computational drug repurposing and precision therapy identification for rare and understudied diseases.

Products and Services: The project delivers three primary deliverables to address inefficiencies in traditional drug development pipelines. First, the awardee constructs a large-scale biomedical knowledge hypergraph that extends the NCATS Biomedical Data Translator's foundational knowledge graph to capture higher-order relationships involving more than two entities—such as drug combination benefits for disease subtypes and patient subgroups. Second, the team develops a reinforcement learning-enhanced large language model (RL-LLM) to extract complex relationships from biomedical literature, incorporating human-in-the-loop quality control with domain expert curation. Third, the project creates HyperGAT, a hypergraph neural network framework incorporating heterogeneous attention mechanisms, interpretability modules, and zero-shot learning capabilities to generate accurate, explainable predictions of drug efficacy. These integrated outputs aim to advance precision medicine approaches and accelerate discovery of synergistic therapies that traditional pairwise-only graph structures cannot effectively represent.