There are 2 ways of analyzing HRV, either from the raw signals or the R to R intervals, in either case, it a fourier transform of the power of the frequencies that is computed. The power is relative, but gives the spectra that reflect either the primary contribution from the sympathetic or parasympathetic influences on the heart and it is the ratio between the two that gives the best evaluation of autonomic balance. The cut off frequencies are dependent on the species, with larger species having lower frequency domains (as would be expected with slower heart rates and respiratory rates as respiratory contributes primarily to the parasympathetic rhythmic slowing of the heart rate).

Dear Douglas,

If 12 lead ECGs are not available and a rhythm strip is the only thing that is available, what HRV domains can be calculated by this strip and how could one calculate the balance between sympathetic and parasympathetic system? The strips are 2 minutes long.

BW

Habib

Dear Gheorghe Zaharia,

To represent interpolated discrete-time sequence of R-R intervals as a frequency spectrum it's nesessary that the researched system was in equilibrium (stationary) in the interval of measurements.

It's evident that in the homeostatic system of autonomic HR control such equilibrium remains unchanged only under the same physical and emotional load (if possible), i.e. during a very short time intervals. Hence

spectral density estimations of HR modulation are meaningful only at these intervals.

As it often happens in the analysis of non-stationary processes, in the HRV standard of measurements... [Circulation, 93: 5, 1996 Mar 1, 1043-65 ] was proposed to use as HRV measures, the time-averaged mean power in the respective frequency bands, calculated at large intervals (intervals of homogeneity). It is known that such time-averaged are not

estimates of the observed process in the statistical sense. Nevertheless, they are relatively steady-state and can be successfully used for comparative analysis.

Best regards,

M.G.

Thank you, Douglas, for your answer. So, the first idea is to apply the Fourier Transform to recorded data and to use the width of the obtained spectra as a measure of HRV. In this case, is it necessary to use a threshold (let’s say, at 10% from the maximum value) in order to obtain the frequency bandwidth? How accurate is this method? The frequency step between 2 frequency adjacent points of the Fourier Transform can reduce the result accuracy? When using the R-R intervals, the obtained result is more accurate? Between these 2 methods, which one is frequently used? I also highly appreciate the Michael Gordon’s remarks concerning the stationaryprocess during the measurements.

Hi Dr. Zaharia,

What you describe is a commonly used HRV index called SDNN (standard deviation of normal-to-normal intervals). It is a good measure of overall HRV capturing both parasympathetic and sympathetic processes. Other indices such as pNN50 (percent of normal-to-normal adjacent intervals greater than 50ms) capture almost entirely parasympathetic processes, as does the high frequency bandwidth from the fast Fourier transform. The paper Dr. Gordan cites outlines the common HRV indices, though some aspects of this paper have since been refuted. For instance, the LF/HF ratio is no longer thought to represent the balance been sympathetic and parasympathetic activation.

I'm not sure what your research application is, but unless you're specifically interested in learning how to calculate HRV indices yourself, I suggest using one of the many programs out there that will analyze your raw data and produce all the HRV indices you'd most likely need. Many suites are free. See this discussion for suggestions: https://www.researchgate.net/post/Is_there_any_free_software_besides_Kubios_in_order_to_analyze_HRV

If you are interested in assessing change in HRV between two recording epochs you might consider the following approach, which is a novel way of gauging shifts in spectral power, Bates M.E., Buckman J.F., Vaschillo E.G., Fonoberov V.A., Fonoberova M., Vaschillo B., Mun E.-Y., Mezić A., Mezić I. The redistribution of power: Neurocardiac signaling, alcohol and gender. PLoS One. 2011;6(12):e28281. dos: 10.1371/journal.pone.0028281. There's a Matlab syntax mentioned in this paper that I'd be happy to share with you.

David

Dear Dr Zaharia,

There are more than 20 HRV indexes grouped under five categories

1- time domain.

2- frequency domain.

3- nonlinear

4- Poincare

5- time-frequency.

Based on your application area, there is an index or several indecies can be more representative.

I would suggest to use HRVAS software to estimate all these indexes

regards

Hilal

Version:1.0 StartHTML:0000000105 EndHTML:0000022156 StartFragment:0000021338 EndFragment:0000022116

You consider HRV as a marker of the activity of the autonomic nervous system. HRV analysis for this purpose is deeply elaborated and even has standards. The results of such analysis are presented in million publications.

We think, more important would be to study and analyze HRV as an active physiological process which continuously modulates blood flow in the body and brain, and, consequently, central and local baroreflexes through baroreceptors in vasculature. Metabolic needs, respiration, and resonances in baroreflex systems are the main generators of the HRV.

Evgeny

You consider HRV as a marker of the activity of the autonomic nervous system. HRV analysis for this purpose is deeply elaborated and even has standards. The results of such analysis are presented in million publications.

              We think, more important would be to study and analyze HRV as an active physiological process which continuously modulates blood flow in the body and brain, and, consequently,  central and local baroreflexes through baroreceptors in vasculature. Metabolic  needs, respiration, and resonances in baroreflex systems are the main generators of the HRV.

Dear Dr. Zaharia,

It is not my habit to promote my own work in these question/answer discussions. But I did write a short, comprehensive review of what one should know when entering the HRV-measurement field. It can be found on the pages 101-106 of my recent review in Physiological Measurement, to be found 'Open Access' on: Article An introduction into autonomic nervous function

Thank you for your answers and the transmitted documents. I am not a medical doctor. My research activities mainly concern radio propagation and digital communications. I participate to a project on the contactless measurement of the cardiopulmonary activity using a continuous wave Doppler radar. The obtained signal is different (if compared to a classical ECG signal) but contains the same information: HR (heartbeat rate), HRV (Heartbeat Rate Variability), respiration rate, etc. For us, it is sufficient to demonstrate the possibility to obtain the same parameters given by an ECG. The medical utilization of these information is not the aim of the project but may allow us a dipper understanding.

The answe was which is the most used parameters. I think these paraneters are the mean and the standard deviation. It does not matter of cardiointervals or instant heart rate values.

Thank you, Valery, for your answer. In fact, I already used these two parametters, but it seems that other parametters are also used by the doctors.

Dear Gheorghe! Yes they are! But more rarely because of their low clinical significance. It's my opinion. This is because of what J. A .Taylor and P. Studinger said in the

Point:Counterpoint: Cardiovascular variability is/is not an index of autonomic control of circulation. J Appl Physiol 101: 676–681, 2006 "...even ignoring the analytic shortcomings and lack of validation that contaminates much of the work in this area, cardiovascular variabilities should not be considered quantitative measures of autonomic outflow owing to their complex and largely undiscovered physiology".

Thank you, Valery, for this information. I am not a medical doctor, my Ph.D. is in "Electronics and Telecommunications". I was involved in a study on the possibility to use a Doppler radar for contactless measurement of HR and HRV. The validation of the proposed system was realized by comparing our results (HR and HRV) with those obtained with a classical ECG. We obtained a very good agreement, even for low power of the transmitted signal. For HRV, it was difficult to find a mathematical definition. This difficulty determined my question.

I would define the heart rate as sequence of time intervals between the excitations of the sinus node of the heart. Hence the variability of the heart rate is the variability of this sequence. The best marker of the sinus node excitation is the P-wave on the ECG. Based on this, you could take a sequence of PP intervals of the ECG. Then you could perform correlation analysis of this time series with the sequence obtained telemetrically. Tight correlation I think will be proof of the relevance of your technique.

Yes, I used the length of the PP intervals in order to determine the variability of the heart rate but I did not consider the correlation between this time series and the sequence obtained telemetrically. I think that the range of the lengths of the PP intervals is quite reduced... However, the correlation coefficient must be high. An intersting idea !