Research Faculty Seminar Series (5/14/24): "Uncovering heart failure phenogroups using clinical data and physiologically-informed machine learning"

Brian Carlson, Ph.D.
Research Associate Professor
Department of Molecular & Integrative Physiology

I will show the efforts in our lab combining human clinical data and computational models of cardiovascular hemodynamics to extract mechanistic parameters describing cardiovascular function in a patient-specific manner. These parameters which now represent the cardiovascular state of each patient in our dataset can then be used to find groupings of like patients using simple unsupervised machine learning tools.  This technique is valuable when trying to understand the finer heterogeneity underlying a broad heart failure diagnosis such as heart failure with preserved ejection fraction.