Neural Network for Forecasting Acute Kidney Injury in Patients With Thermal Skin Burns

The aim of the study was to develop a method for predicting early acute kidney injury (AKI) in patients with thermal skin burns by neural network data analysis. Material and methods. 109 patients were included in the study, 74 of them with thermal skin burns of I-III degree. The levels of cystatin C (Cyst.C), myoglobin (MB), calprotectin (MRP8/14), intercellular adhesion molecules (ICAM, VCAM),my eloperoxidase (MPO) were determined by multiplex analysis on a flow cytofluorimeter. Results. Upon admission to the hospital, generally accepted criteria for the diagnosis of acute kidney injury, such as creatinine concentration and daily diuresis, were determined. The concentration of biomarkers of acute kidney injury, which increases with acute kidney injury, was also determined. In modern medicine, the progressive development of AKI prediction is possible through artificial intelligence. To create a neural network, markers of acute origin with an access point in blood crosslinking were included in the multidisciplinary percepton: Cysta.C; MB; MRP8/14; ICAM; VCAM; MPO. Based on the developed model, the probability of developing AKI in patients with thermal skin burns was determined by analyzing the indicators detected in blood serum, where the accuracy of the forecast of the developed model was 96.3%. Conclusion. The use of a neural network for early diagnosis of acute kidney injury has a high degree of accuracy. This technology should be used.