METHOD FOR NEURAL NETWORK ANALYSIS OF CARDIAL CONDITION Russian patent published in 2012 - IPC G06N5/00 A61B5/402 

FIELD: medicine.

SUBSTANCE: method comprises the stages involving formation of a learning sample in the form of a set (k-1)*m of n-size reference data vectors, construction of decision rules and learning of k*L (L is lead number) neuron networks for analysis of each cardiac k-states in each lead. Then patient's electric cardiac signal is recorded, pre-processed and presented in the form of the n-size vector. The neural network analysis is enabled by comparing the n-size vector to the number (k-1)*m of n-size reference data vectors. The neural network analysis data are used to sample the cardiac k-state with maximum number of signs.

EFFECT: improved algorithm of the neural network analysis of the electric cardiac signal.

4 cl, 11 dwg