Power spectral estimation of short R-R interval segments presenting linear behavior of the SA node

Long segments of heart rate variability signals are usually nonstationary and dynamic, thus limiting the length of the signal available for spectral analysis. An algorithm was developed for accurate power spectrum estimation of short R-R segments not exceeding 15-20 beats. The algorithm was tested during an experiment in two dogs, where frequency modulated pulse train was applied to the vagal nerve after vagal transaction and blockade of the sympathetic system by propranolol. The calculated power spectra presented conspicuous spectral peaks in accordance with the different modulating frequencies between 0.1 Hz and 0.3 Hz. The presence of spurious peaks was negligible even when the analyzed signal segments consisted of 15-20 beats only. These results imply that for a certain range of modulating frequencies the sinoatrial node responds linearly to fluctuations in the parasympathetic tone. System identification methods were used for confirming this hypothesis. For higher frequencies of the modulating signal usually above 0.3 Hz the system tends to deviate from linearity