A novel application of finite Gaussian mixture model (GMM) using real and simulated biomarkers of cardiovascular disease to distinguish adolescents with and without obesity

Abstract Obesity-induced derangements in adipose tissue and other organs lead to the development of cardiovascular disease (CVD). The loss of CV-health in children is a continuum, and the manifestation of overt CVD takes several years. Therefore, robust biomarkers are crucial for its early prediction, prevention, and management. Biomarkers of CVD are highly mutually correlated, and typical regression approaches do not precisely appraise the obesity-induced summative alterations of these overlapping variables. This study examines if the confluence of biomarkers of CVD can distinguish adolescents with obesity from their normal-weight counterparts to illustrate obesity as a strong risk factor of CVD. The biomarkers were measured in a well-controlled study in 21 adolescents. Application of the Gaussian mixture model to these biomarkers identified two distinct groups that matched with the obesity status of participants, which was further confirmed using supervised learning methods. Classification of biomarkers from a simulation study of 1,000 data points, each comprising a vector of five biomarkers and the classification identifier, resulted in two groups that matched with the classification in the simulated dataset. The precise identification of obesity by the pattern of concurring CVD biomarkers in real and simulated datasets confirms obesity as a strong risk factor of CVD.