Federal Grant Award Summary

Duke University's Office of Research Administration received a $446,410 Project Grant from the National Institute of General Medical Sciences (NIGMS) under the Biomedical Research and Research Training program (CFDA 93.859), effective April 1, 2026 through March 31, 2030. The award supports the development and validation of advanced statistical methodologies designed to identify disease subtypes, with specific application to sepsis research. The project delivers three primary products: (1) Bayesian multiview pyramids as an alternative to standard clustering techniques for analyzing heterogeneous patient data; (2) Bayesian multistudy pyramids to enhance reproducibility of disease subtype identification across different clinical studies and patient populations; and (3) robust frameworks for assessing temporal evolution of disease subtypes over time.

These statistical tools and methodologies address significant limitations in current clustering approaches, including lack of robustness and poor reproducibility across datasets. The research combines theoretical development, extensive simulation studies, and validation through application to established sepsis cohorts. While the immediate application focuses on sepsis subtyping to facilitate personalized treatment development and improved patient outcomes, the underlying statistical methods are designed for broad applicability across multiple diseases and clinical syndromes where disease endotyping could inform precision medicine approaches. The work is performed at Duke University's Durham, North Carolina facility.