Sleep and Cardiometabolic Subgroup Discovery and Risk Prediction in UnitedStates Adolescents and Young Adults: A Multi-Study Multi-Domain Analysis ofNHANES and NSRR

Project Summary/Abstract Cardiometabolic (CM) diseases including cardiovascular (CV) and metabolic diseases are the leading cause of preventable death in the United States and Worldwide. CM diseases are interconnected and positively associated with multi-domain Cardiometabolic Risk Factors (CMRFs) such as metabolic dysregulation, obesity, physical inactivity, poor nutrition, and other emerging factors (including and especially sleep disorders). CMRFs are highly and increasingly prevalent in adolescents and young adults, which foreshadows a future epidemic of incident CM diseases as they age. However, existing studies have primarily focused on the adult and senior population with little to no knowledge on the young population. CM data hold great promise to facilitate CM subgroup discovery for early risk stratification and precise prognosis. However, significant gaps exist in fully leveraging CM data. Gap I: Lack of inclusion of multi- domain CMRFs (especially sleep health). Gap II: Lack of “outcome-predictive” CM subgroups in early risk stratification. Gap III: Lack of “subgroup-specific” precise prognosis of “multi-dimensional” CM outcomes. Gap IV: Under-utilization of the rich but “incomplete” multi-domain CM data in NHANES and NSRR. We propose a multi-study multi-domain secondary analysis for CM subgroup discovery and risk prediction in U.S. adolescents (11-18) and young adults (19-39). The objective is to create 2 combined NHANES and NSRR datasets and examine multi-domain CMRFs including metabolic dysregulation, physical inactivity, poor nutrition, and multi- dimensional sleep measures for CM subgroup discovery and risk prediction in the large and diverse U.S. adolescent and young adult population of Hispanics/Latinos, African Americans, Caucasians, and Asian Americans. Aim 1. CM risk subgroup discovery at baseline for U.S. adolescents & young adults. (1.1) Develop a novel sparse Incomplete Multi-domain Mixed-typed Factor Mixture Model (IM2-FMM) for subgroup discovery from incomplete multi-domain mixed-typed CMRFs. (1.2) Apply IM2-FMM to identify, characterize, and evaluate CM subgroups in adolescents and young adults from incomplete multi-domain mixed-typed CMRFs at baseline including: (a) self-reported sleep measures in NHANES; (b) self-reported and objective sleep measures in NSRR. Aim 2. Subgroup-specific prediction of multi-dimensional longitudinal CM outcomes for young adults. (2.1) Develop a novel sparse Transfer Learning-based Generalized Multi-level Model (TL-GMM) to predict multi- dimensional longitudinal CM outcomes from clustered CMRFs at baseline. (2.2) Apply TL-GMM to young adults in NSRR to: (1) examine fixed effects and random effects of baseline CMRFs on CM outcomes; (2) provide subgroup-specific multi-dimensional prognosis of CM health from clustered CMRFs at baseline. Impact: Our study will generate novel insights into CM subgroup discovery to facilitate early and targeted interventions and help establish health promoting behaviors in adolescents and young adults, eventually improving CM health care in their transition to adulthood and reducing CM health disparities and costs as the young population ages.