DEVELOPMENT OF A PROGNOSTIC MODEL TO DETERMINE THE NEED FOR RENAL REPLACEMENT THERAPY IN PATIENTS AFTER CORONARY ARTERY BYPASS GRAFTING

Aortocoronary bypass surgery using artificial circulation is often accompanied by the development of acute renal injury, especially in the presence of chronic kidney disease in patients. Early application of renal replacement therapy methods is aimed at reducing postoperative complications. In this regard, there is an active search for biological markers that allow for the timely detection of this complication. Aim. Development of a predictive model aimed at identifying the need for renal replacement therapy in patients with coronary heart disease after coronary artery bypass grafting based on established biomarkers. Materials and Methods. The data sample consisted of 120 patients of both sexes with coronary heart disease before and after coronary artery bypass grafting and 30 practically healthy people of comparable age and sex. Retrospectively, the patients were divided into groups according to the nature of the course of pathological changes in the kidneys (uncomplicated, complicated) in the early postoperative period. Cytokine levels, matrix metalloproteinases and their inhibitors were studied, obtained from venous blood serum and urine taken a day before surgery, on the 1st, 2nd and 7th days after surgery. Statistical evaluation of differences in indicators in groups was carried out using nonparametric Wilcoxon, Mann-Whitney and Kruskal-Wallis tests. Results. Artificial neural networks with a single-level architecture were used to build a predictive model. As biomarkers of renal complications after coronary artery bypass grafting, indicators were identified whose median values differed statistically significantly in the groups of patients before surgery and in the postoperative periods, as well as with the control group: TGF-α, TGF-β1, TGF-β3, IL-6, MMP-1, MMP-7, MMP-8, TIMP-2, TIMP-3 in blood serum; TGF-α, TGF-β1, TGF-β3, MMP-1, MMP-7, MMP-8, TIMP-3 in urine. Based on these indicators, taken on the second day after surgery, a predictive neural network model consisting of five neurons in one layer was developed. In the test sample, the model predicts 91.7% of the time that renal replacement therapy will not be necessary in patients undergoing coronary artery bypass grafting and 100% of the time that it will be necessary. The quality of the model on the test sample is 93.33%. Conclusion. The study makes it possible to adjust treatment methods for patients with coronary heart disease after surgery at an early stage in order to improve the prognosis and enhance their quality of life. The conducted research makes it possible to adjust the treatment methods of patients with coronary heart disease after surgery at an early stage in order to improve prognosis and enhance their quality of life.