Research on patterns in blood pressure dynamics based on approximation of experimental data with the help of criteria of smoothness

Abstract

To reproduce the trends of the studied process, it is necessary to reduce the variability of output data, i.e. to smooth, to draw a smoother, more waveless curve, which should be as close to the experimental points as possible. To improve the approximation efficiency of experimental data, it was suggested to use complex criteria, including minimizing the mean square error of approximation while using conditions of smoothness of an approximating curve based on the analysis of second derivatives. The suggested approach was compared with the method of moving average, which was widely used for smoothing processes under conditions of a priori uncertainty. The proposed methods were used to smooth the monitoring data of the blood pressure. It was shown that the optimal smoothed dependencies ensure sufficient smoothness to highlight short-term fluctuations in the average blood pressure, in order to identify the persistent patterns of their dynamics, which was important for predicting the human health state.