Open AccessEstimation of human metabolic age using regression and neural network analysis
Author(s) -
O. V. Korkushko,
A. V. Pysaruk,
V. P. Chyzhova
Publication year - 2021
Publication title -
zaporožskij medicinskij žurnal
Language(s) - English
Resource type - Journals
eISSN - 2310-1210
pISSN - 2306-4145
DOI - 10.14739/2310-1210.2021.1.224883
Subject(s) - regression analysis , artificial neural network , medicine , linear regression , metabolic syndrome , creatinine , regression , cholesterol , diabetes mellitus , endocrinology , statistics , mathematics , computer science , machine learning
The aim is to develop the methods for assessing the rate of human aging by metabolic parameters (metabolic age).Materials and methods. The study examined 120 subjects aged 40–80 years. All the people included in the study underwent the determination of plasma glucose concentration, lipid profile – total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein, very low-density lipoprotein cholesterol and creatinine as well as the standard glucose tolerance test. Validation of the panel of indicators was carried out using regression and neural network analysis.Results. According to the study results, the standard error in determining the metabolic age using the multiple regression equation was 9.31 years, and using the neural network – 3.18 years.Conclusions. The methods that we have developed for assessing the rate of metabolic aging showed sufficient (regression analysis) and high (neural network analysis) accuracy and can be used to assess the risk of metabolic syndrome, cardiovascular disease, and type II diabetes. The implementation of the proposed methods would not only identify people at risk for pathology, but also assess the effectiveness of treatment, prevention and rehabilitation measures.