Prognostic mathematical models of the development of chronic non-viral liver diseases

Abstract

Chronic non-viral liver diseases are one of the most pressing problems of modern internal medicine. Over the past two decades, there has been a steady increase in diseases of this group due to the widespread prevalence of major risk factors, such as obesity, diabetes mellitus, physical inactivity, alcohol abuse, irrational use of medications, etc. Since most of the patients represent the most able-bodied and active part of the population, this circumstance inevitably leads to significant financial, economic and humanitarian losses, which gives the problem acute social significance. The purpose of the study. To develop mathematical prognostic models of the development of the most common chronic non-viral liver pathology using the most important predictors of these diseases. Material and methods. The contingent included in the study consisted of 412 young and middle-aged men, 138 of them with non-alcoholic fatty liver disease, 50 with alcoholic liver disease, 157 with alcoholic-metabolic fatty liver disease and 67 people without pathology. The subjects were measured anthropometric indicators (height, body weight, calculation of body mass index, waist circumference), determination of the type of alcohol consumption according to the AUDIT questionnaire, the level of physical activity according to the IPAQ questionnaire. In the biochemical study of blood serum, the parameters of alanine and aspartate aminotransferase, total and direct bilirubin, alkaline phosphatase, gamma-glutamyltranspeptidase, glucose, insulin, and lipid spectrum were determined. The study of the qualitative and quantitative composition of the intestinal microbiome was carried out by sowing feces on nutrient media, followed by the determination of the species of microorganisms and the assessment of their quantitative content using microscopy. Ultrasound examination of the hepatobiliary zone included measuring the main sizes of the right and left lobes of the liver, as well as identifying the main signs of diffuse liver damage. To assess steatosis and liver fibrosis, a FibroScan 530 device was used with the functions of evaluating the parameter of controlled attenuation of ultrasound and transient elastography by controlled vibration. Statistical processing and analysis of the results obtained were performed using the StatTech 2.8.8 program. Results. Mathematical prognostic models have been developed to determine the probability of developing non-alcoholic fatty liver disease, alcoholic fatty liver disease and alcoholic-metabolic fatty liver disease. All the regression models obtained were statistically significant with sensitivity and specificity of more than 90%. AUROC diagnostic accuracy values exceeded 0.9 units, which characterizes the quality of diagnostic models as excellent. Conclusions. The use of mathematical prognostic models using publicly available data from objective and laboratory research methods allows optimizing the diagnosis of chronic non-viral liver diseases.