O.D. Yuryeva

An increasing of human resistance to effects of stress is one of the actual tasks of man-operators - training. The method of heart rate variability analysis based on processing of RR-intervals (time intervals between adjacent heart beats) is of frequent use in human psychophysiologic condition prediction. The signal formed by the series of RR-intervals is nonstationary. The most well known methods of nonstationary signal analysis are algorithms, based on the use of parametric modeling of processes, correlation and spectral analysis. These approaches make it possible to detect the signal sections that can be considered as locally stationary. But they give no possibility of assessing of the signal nonstationarity range. The results of experimental examination of different parameters characterizing cardiac rhythm signal nonstationarity and also combinations of these parameters are presented for purposes of increasing of heart rate variability parameter reliability and estimation of human psychophysiologic condition in the process of psychophysiologic testing and professional work. The signal nonstationarity parameters were calculated using two running windows. The following parameters were considered for assessing of signal statistical characteristic changes: mean value, standard deviation and linear trend slope angle. The module of difference between values calculated for the first and the second windows were analyzed for indices "mean value" and "standard deviation". In order to estimate signal spectral characteristic changes autoregressive model prediction error was selected. Both separate cardiac rhythm signal nonstationarity parameters and different parameter combinations were studied. Four methods for combining the parameters were used: method 1 - square root from sum of squares of parameter values; method 2 - sum of parameters; method 3 - sum of parameter squares; method 4 - product of parameters. The minimum ratio value between mean values of indices estimated at stationary and nonstationary fragments was considered as an criterion for the best method of nonstationarity integral index calculation. It was shown that the best results of nonstationarity assessment may be obtained when using the combination of the parameters "mean value", "standard deviation", "linear trend slope angle" and "autoregressive model prediction error" by product method. The nonstationarity integral index can be used for assessing of the adequacy of the subject-s response in the process of psychophysiologic testing and professional work and also for controlling of the training success.