Intellectual Systems with Virtual Flows in Predicting Cardiovascular Complications

To predict cardiovascular complications, it is proposed to use “weak” classifiers with virtual flows. The “weak” classifier is based on studies of electrical conductivity at biologically active points (BAP). The action is formed in the form of premises in BAP voltage pulses of increasing amplitude. As a reaction, the voltage measured is used on the current resistor after the transient attenuation, connected in series to the human body. The classifier is built in the form of a hybrid model and includes a neural network and a virtual channel that forms an additional informative feature. The virtual channel is formed on the basis of a priori given functional dependence on electrical conductivities of the biomaterial calculated in each test pulse. The exponential model is investigated as a functional dependence. Hybrid classifiers of individual risk and group risk are taken into consideration. Experimental studies of the effectiveness of classifiers with virtual flows were carried out on patients with myocardial infarction in remission. The article presents the comparative characteristics of quality of indicators of myocardial infarction relapse prediction for classifiers using traditional informative features, classifiers with additional weak classifiers performed on the basis of bio impedance studies, and with additional classifiers performed on the basis of bio impedance studies using virtual flows.