Comparative analysis of mathematical simulating of the non-stationary heart rate variability in diagnostic medical systems

Reliable mathematical procedures and software systems to carry out the automatic cardiac rhythm regulation which include recording of rhythmogram signal, processing and formation of diagnostic conclusions with subsequent correction of patient's condition, are the important part of medical cyber-physical systems (MCPS). To detect and correct the arrhythmias of a different nature at early stages, we are to consider the problems of mathematical simulation of the non-stationary heart rate variability (NHRV). In contrast to the traditional simulating of a rhythmogram as an amplitude-modulated signal (AMR), we propose the model of a rhythmogram as a frequency-modulated signal (FMR) having the form of Gaussian peaks of equal height with the centers located on the irregular grid of time. We determined the quantitative spectral parameters of FMR by the double continuous wavelet transform (DCWT). Within the framework of the traditional approach, we calculated these parameters using the fast Fourier transform (FFT). We compared the FMR and AMR approaches to NHRV and detected the difference in spectral properties calculated on the basis of these two models. The deviations from the true values of spectral parameters in the AMR approach were especially evident when NHRV had a noticeable trend throughout the total period of observation. The possibilities to apply the proposed FMR concept for continuous real-time diagnosis of the cardiovascular system under load, for example, during various functional tests, are discussed.