METHOD FOR DETERMINING RISK OF HEART TRANSPLANT REJECTION Russian patent published in 2018 - IPC A61B8/00 G06N3/02 

FIELD: medicine.

SUBSTANCE: invention relates to medicine, namely, in transplantology and cardiology, and can be used to determine the degree of risk of transplant rejection. Method involves identifying predictors. Using an ultrasound method of investigation, speckle-tracking echocardiography, reveal a complex of 13 predictors of cardiac transplant rejection: global peak systolic strain of left ventricular (GLSLV, -%), longitudinal left ventricular strain in 4 chamber position (A4C, -%), longitudinal left ventricular strain in a two-chamber position (A2C, -%), longitudinal left ventricular strain in three chamber position (A3C, -%), global peak systolic strain rate of left ventricle), (GLSTRLV, -c-'), radial strain of left ventricular (RadSLV, %), radial strain rate of left ventricular, (RadSTRLV, c-'), circular strain of left ventricular (CirSLV, -%), circular strain rate of left ventricular (Cir STR LV, -c-'), twisting (twist, ''), rotation of the apical segments of the left ventricle), (ROT APEX, °), rotation of basal segments of the left ventricle (ROT BASE, °), rotation of the middle segments of the left ventricle (rotation of the middle segments of the left ventricle ROT MID, °). Then, using the formula, using computer analysis, we calculate the risk of rejection of the cardiac transplant: where Z k – “output” data of the third layer for 4 groups, e is an exponent, i 1…8 – index of the location of the “output” data of the second layer, the neural network, j 1…13 – index of the location of the “output” data of the first layer of the neural network, X'' – “output” data of the first layer of the neural network, w j – weight 13 of the normalized values, w I – weight 8 of normalized values, e – thresholds. Formula combines three layers of a neural network: transform 13 predictors into “normalized” values of the first layer of the neural network and get 13 “normalized” values (j). Transform the values of the first layer of the neural network into the second layer of the neural network and 8 “normalized” values (7) are obtained at the “output”. Transform the parameters of the second layer of the neural network according to the indicated mathematical formula into the “output” data of the third layer of the neural network, which includes: rejection group 1 – recipients without cellular and humoral rejection, ACR 0, AMR 0; group of rejection 2 – recipients with cellular rejection of 1 degree, ACR 1; group of rejection 3 – recipients with cellular rejection of 2 degrees, ACR 2; group of rejection 4 – recipients with humoral rejection of 1 or 2 degree, AMR 1. Then get Zk from 0 to 1, corresponding to the risk of rejection of the cardiac graft. In this case, the recipient will belong to that group in the third layer, whose values are modulo 0.9–1.

EFFECT: method allows to determine the risk of rejection of the cardiac transplant, to reveal early predictors of cardiac transplant rejection during the ultrasound stage, to identify the degree of risk of rejection of the cardiac graft, to classify the recipients of rejection of the cardiac graft by the mechanism and degree of rejection.

1 cl, 15 dwg, 4 ex