Project Summary. Unhealthy sleep can manifest in multi-organ associations, affecting cardiometabolic, immune, behavioral, cognitive and other systems. Research has pointed to the predictive and diagnostic potential of biomarkers derived from physiological sleep signals, not just for sleep disorders but also for a broad range of other diseases. However, although sleep is a dynamic process, that changes over the night, the typical metrics considered as possible biomarkers fail to capture this directly. Based on polysomnography data collected on over 13,000 participants from NHLBI National Sleep Research Resource (NSRR), we will develop a framework for quantifying individual differences in the dynamic structure of cardiac physiology during sleep. We will study the relationships of derived phenotypes with cardiometabolic disease state as well as its genetic diathesis, to better characterize the causal links between sleep-related physiology and health. Specifically, we will apply time series clustering methods to capture individual differences in how core physiologic parameters vary across the night, focusing on electrocardiographic signals. We will then evaluate these derived traits, in terms of reliability, heritability, demographic and chronotype associations, and prediction of cardiometabolic disease with a focus on hypertension and diabetes. Secondly, we will use genetic approaches to study the causal relationships between sleep-related dynamics and cardiometabolic disease, by developing models that use polygenic risk scores from external genome-wide association studies. In addition to generating new methods and tools for large-scale analyses of physiologic sleep signal data, this work has the potential to discover novel biomarkers for cardiometabolic disease, which can inform risk stratification models and point to therapeutic targets.