Title: Informed Clustering of Multi-View Data
Seminar Speaker: Amy Herring, Sara & Charles Ayres Distinguished Professor of Statistical Science, Global Health, and Biostatistics and Bioinformatics at Duke University.
Sepsis is a life-threatening condition characterized by a dysregulated host response to infection. A growing body of research seeks to identify clinically meaningful sepsis subtypes by clustering patients based on demographic, clinical, and biological measurements, with the goal of informing personalized treatment strategies. However, conventional clustering approaches suffer from sensitivity to small differences across datasets, limited clinical interpretability, and poor reproducibility across study populations. We develop new Bayesian methods to address these challenges. First, we introduce clinically interpretable Bayesian clustering models that explicitly quantify uncertainty and improve robustness across heterogeneous datasets. Second, we propose a Bayesian factor modeling framework for integrating high-dimensional genomic data with clinical variables. The approach incorporates prior biological knowledge by leveraging gene pathway information to guide the latent structure, improving interpretability and stability of the inferred factors. We illustrate these methods through an application to sepsis patient data, demonstrating how incorporating prior biological knowledge and clinically meaningful structure can improve the identification of reproducible disease subtypes. More broadly, the proposed framework provides a principled statistical approach for subtype discovery in complex biomedical settings involving heterogeneous clinical and genomic data. Based on work with Alex Dombowsky, David Dunson, Deng Madut, Lorenzo Mauri, Cameron Miller, Matt Rubach, Federica Stolf, and Maoran Xu.
Register here for the seminar.
Register here for streaming.
Note that a post-lecture reception will be held in Thompson Library in the Mortar Board Room 202 from 4-5 PM.