Prospects for the use of artificial intelligence technologies in predicting the risk of cardiovascular disease (CVD) outcomes in patients with type 2 diabetes mellitus
Abstract
Abstract: The study investigates the role of COVID-19 in anamnesis and other features on mortality in 114 patients with type 2 diabetes and cardiovascular diseases during their two-year observation. We performed statistical comparison of two groups of patients and found differences in distributions of some features such as Creatinine, Glomerular Filtration Rate, Total cholesterol, Uric acid, Natriuretic peptide, Ejection fraction, Atrial diastolic flow rate, Isovolumic relaxation time, and others. We also used Logistic Regression and 4-fold crossvalidation to reduce feature space. We built the optimal model using Logistic Regression to predict mortality in two years after medical examination with Concordance Index = 0.955 on test set and Concordance Index = 0.987 on training set.