Innovative deep learning algorithms for predicting arterial stiffness in diabetes patients

Abstract

Chronic diabetes mellitus is defined by an abnormality in the metabolism of proteins, lipids, and carbs. Diabetes is linked to both Arterial Stiffness (AS) and hypertension. The capacity of AS to predict diabetes is also superior to that of hypertension. However, the research topic has not yet been fully explored, and the preliminary findings are not very promising from the literature. An early illness diagnosis will help save lives and hence is essential. Due to its significant potential for creating effective tools for risk prediction, diagnosis, treatment, and the management of various disorders, Machine Learning (ML) has become quite popular among medical professionals. Brachial-ankle Pulse Wave Velocity (baPWV) is a special indicator of systemic arterial stiffness used in clinical settings. Although significant efforts have been made to lower the risk of Cardiovascular Disease (CVD), a huge number of people have elevated arterial stiffness risk for the disease, and the baPWV measurement is only occasionally used. As a result, a straightforward and practical clinical technique is required to evaluate high arterial stiffness in routine clinical practice. A crucial step toward the monitoring and diagnosis of high arterial stiffness is the creation of a risk rating system based on predictors, i.e., clinical data. This is the first problem that have been taken up in this research. Least Absolute Shrinkage and Selection Operator (LASSO) regression is used to choose the biological parameters of the diabetic patient that are the most important critical aspects. Gaussian regression with Cat optimization will select the top five characteristics from the selected features, such as age, Body Mass Index (BMI), Fasting Blood Glucose (FBG), Systolic Blood Pressure (SBP), and Diastolic Blood Pressure (DBP), to produce an accurate classification of the arterial stiffness risk factors. newline