Neural network modeling of medications impact on the pressure of a patient with arterial hypertension

Abstract

The problem of individual blood pressure prediction and control of hypertensive patient is addressed. The method of predictive model synthesis is proposed. It uses the windows method to form training sample from the original data, the feature selection based on information criterion in discretized feature space, the instance selection based on transformation of original multi-dimensional feature space into one-dimensional space of generalized axis, and the multi-layer feedforward neural network trained by the Levenberg-Marquardt method. On the basis of obtained model the proposed method provide selection of optimal individual medications combination. The software implementing the proposed method is developed. The computational experiments on the model synthesis are conducted. The dependencies between the method parameters are experimentally obtained. The recommendations on assignment of method parameters are given.