The decrease of complexity of the RR variability resulting from the breathing slowing is a consequence of the sinus node responsiveness, while the decrease of RR variability complexity due to aging and pathology is a sign of the reduction of control mechanisms able to govern cardiac rhythm provided that breathing rate is controlled.

Dear Debanjan,

You raised an interesting discussion. In fact, we can find contrasting conclusions about changes in complexity and I believe this is due mainly to what we understand by "complexity". For example, some researchers consider that entropy is a measure of complexity, so that the higher the signal's entropy, the higher is the complexity of the system that generated that signal. In contrast, other researchers believe that entropy, by itself, is not a direct measure of complexity. For those researchers, entropy can be used used as part of the process that characterizes the complexity (such as in multiscale entropy and entropy of entropy), but not directly. In addition to entropy, other signal descriptors are often used to characterize the system's complexity, such as the fractal structure of signals (detrended fluctuation analysis, fractal dimensions) or asymmetry properties.

So, the question here is: what are the features of signals necessary to characterize its complexity? Since this is still an open question, this understand may vary among the researchers, leading to those contrasting conclusions.

Best regards,

Luiz

Hi, @Luiz I have found this resource https://archive.physionet.org/tutorials/cv/. It describes how complexity and variability are fundamentally different.

Hi Debanjan,

Yes, this tutorial is excellent! Very informative and didactic. Thanks for remembering it.

Best

It is well-known that the complexity of RR variability series decreases while slowing breathing (A. Porta et al, Med Biol Eng Comput, 38, 180-188, 2000) and this effect is likely to be the consequence of the low-pass characteristic of the sinus node transfer function making less evident irregularities of the respiratory rhythm occurring at higher rates whether the dominant frequency of the respiratory sinus arrhythmia decreases (J.P. Saul, et al, Am. J. Physiol. 261, H1231-H1245, 1991; A. Porta et al, Med Biol Eng Comput, 41, 52-61, 2003). This effect is largely underestimated in literature when interpreting the reduction of RR variability complexity as a consequence of other influences (e.g. aging and pathology).

Alberto Porta Thanks, So, this reduction in complexity of cardiac rhythm is beneficial for humans or not? As we read in literature that HRV biofeedback has health benefits and used in cardiac rehabilitation.

The decrease of complexity of the RR variability resulting from the breathing slowing is a consequence of the sinus node responsiveness, while the decrease of RR variability complexity due to aging and pathology is a sign of the reduction of control mechanisms able to govern cardiac rhythm provided that breathing rate is controlled.

I think this is not a good approach to target a biomarker instead of the pathophysiology to treat/manage a disease, especially when the biomarker is non-specific. The problem with cardiac autonomic indexes (e.g. HRV) is that they are non-specific and are influenced by a lot of pathological and physiological conditions at different central and peripheral levels. An intervention that shortly influences these indexes does not necessarily will help to treat/manage the disease. We always need to first ask what is the pathophysiology of the disease and through what mechanisms the intervention can influence it? if there are plausible mechanisms then we need to conduct well-designed mechanistic studies to test the efficacy and mechanisms of the intervention. Starting from the biomarker, which is just a marker not the target, can be misleading.

I see your point, Ali. HRV is affected by influences other than the autonomic nervous system (e.g. intrinsic pacemaker cell mechanisms) and I think that all those influences are variable, depending on the study protocol and population conditions. Moreover, one of the biggest limitations of nonlinear HRV indices is the lack of a direct relationship with specific physiological mechanisms. Instead, they provide information on global control, the complex of the system as a whole, reflected in the dynamics of heart rate fluctuations.

I aggree to Ali Gholamrezaei. Cardiovascular and especially psychiatric diseases and their individual development over years to decades are caused multifactorial. Therefore HRV parameters should not be used as a marker to monitor a treatment neither as an individual predictive prognosis.

On the other hand, I think we cannot disregard the link between HRV and autonomic modulation. We have to bear in mind that the autonomic system reflects the homeostasis of the whole organism, and so, can be altered in many conditions, including non-cardiovascular diseases.

In an interesting cross-talk debated published by J. Physiol., Malik et. al. emphasize the importance of HRV as a tool to measure autonomic responsiveness instead of autonomic tone. It is worth reading (https://physoc.onlinelibrary.wiley.com/toc/14697793/2019/597/10).

Luiz Eduardo Virgilio Silva Thanks, Mark Boyett concludes in his paper - " In summary, HRV is primarily determined by heart rate and cannot be used in any simple manner to determine cardiac autonomic tone". Whereas, Marek Malik says, -" HRV offers a measure of autonomic responsiveness which is clinically very useful in the diagnosis and detection of autonomopathies and neuropathies. " To me a lack of consensus among researchers confuses the new researchers. Why can't we have one study that will just prove or disprove whether HRV measures autonomic tone or not?

Debanjan Borthakur, have you ever tried breathing at 0.1 Hz (and recorded an ecg at the same time)? It's impressive to see the amplitude of the 0.1 Hz oscillations in the RR intervals caused by the synchronization of heart rate oscillations and baroreflex/blood pressure oscillations mediated by slow-paced breathing. A single performance of slow-paced breathing for e.g. 10 to 15 minutes might have no effect on cardiovascular control. However, a recent study showed that a 30-day intervention of slow-paced breathing for 15 minutes per day improved vagal activity even in young healthy subjects (they should have had good vagal activity anyway) - see:

Article Influence of a 30-Day Slow-Paced Breathing Intervention Comp...

Article Slow Breathing Improves Arterial Baroreflex Sensitivity and ...

Hence, I hypothesize that patients with hypertension may support their 'classic' treatment by regurlaly exercising slow-paced breathing because it is some kind of 'training' for baroreflex function and blood pressure control.

Dirk Cysarz Hi, Yes. I do .1 hz breathing regularly. I actually wrote a Thesis on it. Please have a look at and please let me know your thoughts https://macsphere.mcmaster.ca/handle/11375/25887

Debanjan Borthakur thank you for bringing up such an interesting discussion. Still, I find it difficult to follow the discussion about HR complexity in general terms and its changes during slow breathing. Thank you also for bringing up our paper

Matić Z et al., 2020

Article Breathing frequency and body posture induced perturbations i...

where we reported that type of complexity pattern (i.e. self-similarity or irregularity) and scale (short scale or long scale) are decisive factors of the type of HR complexity change during 0.1Hz breathing. Both patterns and scales are reciprocally regulated in standing with 0.1Hz breathing. We also reported specific pattern of nonlinear cardiorespiratory coupling specific for standing with 0.1Hz breathing. This pattern was not present for supine position with 0.1Hz breathing. So, there must be something more but sinoatrial node transfer function. We agree with Alberto Porta that SA node transfer function plays its role, but higher, hierarchically organized control mechanisms regulate specific HR complexity patterns, tightly correlated to the complexity of behavioral task.

Looking forward to your feedbacks

Thanks Tijana Bojić . For supine and standing in both conditions 0.1 Hz breathing will lower the complexity of the HR but as you mentioned complexity of behavioral task will have an effect on it. I assume that reduction in HR complexity during slow breathing is actually increasing the reserves ( Arpaia and Andersen 2019 ) and will eventually tend to reduce the sympathetic dominance over time so it's not representing a pathological state !

Debanjan Borthakur I would like to bring your attention to the concluding paragraph of the Matić Z et al., 2020 Introduction:

"Finally, as it goes for the simplest non-linear systems, RRI and respiratory regulation in coupled behavioral states like supination with slow 0.1 Hz breathing (supin01) and standing with slow 0.1 Hz breathing (stand01), most probably contravene the principles of proportionality and superposition (Goldberger, 2006). Slow 0.1 Hz breathing in two specific body postures could potentially have completely different effects on cardiorespiratory complexity parameters with respect to the predicted simple summation. "

Strongly recommend the cited paper:

Goldberger, A. L. (2006). Complex systems. Proc. Am. Thorac. Soc. 3, 467–472. doi: 10.1513/pats.200603-028MS

Debanjan Borthakur Thanks for sharing your thesis, very interesting indeed. However I cannot agree with your conclusion that " slow breathing had a larger effect than music on physiological measures of stress such as in HRV " since there was no diff btw interventions during baseline2. Diff in HRV during intervention cannot be called "physiological measures of stress" since it is in response to intervention (breathing vs music) not stress. To see whether intervention can counteract the effect of stress you need to measure the response after intervention as you did during baseline2 and there was no diff in HRV there.

Also, I wouldn't take SDNN as a cardiac PNS measure, as you also mentioned, it is influenced by both cardiac SNS and PNS activity, RMSSD is a better choice and I'm surprised why you didn't analyse or report it?

What is also interesting in your study is an increase in HR during slow breathing which is similar to some other reports, including ours, and seems to be due to the effect of cognitive demand for controlling the breathing, vs relaxation. This indeed indicates a bit higher cardiac sympathetic tone or lower cardiac parasympathetic tone during slow breathing, not to confuse HRV (cardiac autonomic modulation) with cardiac autonomic tone. Also see

Article Heart rate variability: Origins, methods, and interpretive caveats

Another very important point is that even though HRV was modulated during slow breathing (which is not surprising) that doesn't mean the mechanism of the effect of slow breathing on stress response is mediated by HRV. Was there any association between HRV in response to slow breathing and subjective stress responses? You can statistically test this by mediation analysis, and I expect no mediation. We found similar results with pain.

Article Respiratory Hypoalgesia? The Effect of Slow Deep Breathing o...

Article Can Slow Deep Breathing Reduce Pain? An Experimental Study E...

Ali Gholamrezaei Thanks for your comments. I feel that we did not see differences in the baseline2 because subjects were not familer with the protocol and maybe 5 minute was a too short duration. Lehrer's protocol is 20 minutes twice a day and If I remember correctely they have found that long term practise of slow breathing increases baroreflex sensitivity and resting HRV. What do you think? subjective stress responses was only correlated with galvanic skin response. Most of Lehrer's work used SDNN , so I considered SDNN. Thanks for your papers. What do you think of the complexity analysis I did in my Thesis?

Ali Gholamrezaei, I did not know that slow breathing could increase HR. Interesting your point that it could be the effect of the cognitive demand for controlling breathing. Do you think that this cognitive demand could be responsible for directly (centrally) eliciting a higher sympathetic activity, and thus, increasing HR? Moreover, if the increase in HR is a consequence of the higher sympathetic activation, should not we also expect an increase in AP, in contrast to the findings showing that slow breathing attenuates AP?

Thank you!

Debanjan Borthakur I suggest always looking at the first hand evidence. See the following original studies on the effect of ANS blockade on HRV indexes:

Table 1 in this article

Article Influence of Atropine on Fractal and Complexity Measures of ...

Table 2 in this article

Article Effect of autonomic blockers on heart period variability in ...

and see how HRV indexes alter after ANS blockade, this will help to better interpret the indexes. So studies that have used ANS blockade either pharmacologically or anatomically are the evidence we should look for to interpret physiological mechanism behind HRV and alteration in HRV. Even, evidence in a specific context (e.g. resting state) is not generalisable to other context (e.g. stress test, ANS challenge, etc.).

SDNN reflects all the variability in a RR-interval time series (similar to total power) and therefore it can not distinguish btw PNS and SNS components. RMSSD in contrast reflects (rapid) beat to beat variation and reflects mainly PNS (read the paper I shared above about the time delay in SNS and PNS systems).

RMSSD and SDNN are however highly correlated, but that doesn't mean they are the same.

I have no experience with complexity analysis, again you need to look at the evidence that have used a valid method for understanding the physiological mechanisms. You can create hundreds of new indexes from HRV measures, adding, dividing, multiplying, etc. but at the end all of these should be validated against a gold standard in a specific context.

Luiz Eduardo Virgilio Silva The increase in HR during controlled breathing we have seen in few experiments (not all experiments) is very small, about 2 bpm, and we think it is due to cognitive demand because we have seen it similarly in both controlled breathing at normal (~.23 Hz) and slow (.1 Hz) frequencies vs. uncontrolled breathing.

The mechanism for the effect of slow deep breathing on blood pressure is not known, mostly is observed in long-term practice, not short term single session (at least in our experiments), and can be due to both psychological (relaxation) and direct physiological (central ANS, baroreflex) modulations. Studies also have been different in methodology;

- sitting vs supine position

- depth of breathing and possibility of hyperventilation

- duration of breathing

- relaxing vs attentive

So we don't know and we cannot directly compare studies. The best would be to run further studies and measure SNS by MSNA for example during controlled breathing at various frequencies and under diff body postures with relaxing and attentive suggestions to see the effect of each of these important components.

I have been following the beat to beat heart rate of patients using Thought Technology BioComp and index finger PPG since ~2000. I have also been following the literature rather obsessively since the mid 1990's. My sense is that most of what is being written about regarding clinical uses of HRV is more akin to mythology than science. HRV is affected by far too many confounding factors to have any predictive value except for highly specified conditions in highly constrained contexts. Posture, respiration rate, respiration volume, respiration flow, movement, oxygen demand, airway resistance, abdominal distress, anxiety, depression, PTSD, addiction, eating disorders, pain, immune status, medications, caffeine, nutritional status, ... all have been reported to affect HRV. When everything affects HRV then HRV is a marker for nothing. For a clinical test to be useful one needs to know the specificity and sensitivity of the test under the conditions being tested and in the population being tested. I have seen very few articles that report the sensitivity and specificity of any HRV metric for diagnosing or ruling out any condition.

Specific HRV metrics are useful in highlighting an increased risk when very specific questions are being asked and the context is tightly constrained: for example, in predicting neonatal sepsis in the NICU, or in predicting a subsequent adverse cardiac event in a person who has had a recent cardiac event. However, those examples do not mean that those specific HRV metrics are a marker for anything other than those phenomena in those contexts. Nor do those examples mean that other HRV metrics are markers for anything.

What seems to happen is that a data set shows a regression line with non-zero slope between an HRV metric and some condition with p

Thank you Joseph Arpaia for the comment.

If that could be of any interest, in my profile there is the summary of a dataset of 4 months of EKG in real life. The dataset is freely avaliable.

That would be of interest. Thank you so much for doing that work and making the data available. I have worn various monitoring devices for extended periods of time and that can be quite uncomfortable and takes a lot of effort. So again, thank you very much.

In my opinion, the real magic of ~0.1 Hz breathing is the maximal amplitude in RSA (c.f. HRV biofeedback or "resonance frequency breathing"). This means that each breath cycle is creating a strong autonomic pendulation, i.e. a phase-coordinated swinging between sympathetic (neurovascular) and parasympathetic (cardiovagal) branches of the baroreflex. Such neurovascular-bioacoustic resonance is the result of the harmonic eigenfrequency of the baroreflex system [c.f. Vaschillo et al.] being stimulated by additive reinforcement (or "constructive interference" to use physics terminology) with the input of the breath (i.e. intrathoracic pressure variations causing RSA). As an analogy, imagine blowing across the opening of a glass bottle in the right way and the whole bottle will sing as a Helmholtz resonator. Thinking too much about complexity is a sideways direction in my opinion; the juicy bit to be recognized is the synergetic "autonomic fusion", where sympathetic and parasympathetic reinforce each other continuously in a spiral. (Matić Z et al., 2020 mentions this as "joint physiological enhancement/synergy of sympathetic and vagal modulation"). Another analogy: Like the harmonic motion of a swinging pendulum, in which potential and kinetic energy are exchanged, so is the energy exchanged between sympathetic (neurovascular constriction) and parasympathetic (cardiovagal braking). It can also be viewed as complex phasor: with a rotating vector around the unit circle, the projection onto the x axis forms the cosine function, and onto the y axis makes the sine function. The centrifugal magnitude has a life-enhancing effect, generating a maximum of energy output with a minimum of energy input (another definition of resonance). Another metaphor: charging the resilience battery (c.f. Laborde's Vagal Tank Theory) with a generative "turning of the crank" by the respiratory diaphgram.

Going even further out on a hypothetical limb, I invoke the work of Tong et al. (2015). "Can apparent resting state connectivity arise from systemic fluctuations?" with regard to 0.1 Hz as a primal frequency related to ontogenesis. The metaphor here is the sand patterns on a Chladni plate, and a Zen garden. Someone steps all over the patterns in the sand, creating pathological distortions. Ringing the "primordial bell" of the intrinsic 0.1 Hz eigenfrequency could reorganize the distortions from within, by fundamental resonance. The work of O’Callaghan et al. (2016 and 2019) supports this thesis, demonstrating the promising effects of enhancing RSA on cardiac neurogenesis.

I published some thoughts on this issue in the article "Physiological interpretation of heart rate variability spectral analysis data" ISSN 0201-8489 Фізіол. журн., 2019, Т. 65, № 2 P. 31-42. https://fz.kiev.ua/journals/2019_V.65/2019-2/2019-2-31-42.pdf

Dear Mykola,

Thank you very much for sharing this paper. It is an interesting and detailed report on potential sources of components of HRV in linear domain. It surely enriches the comprehension of peripheral information sources. Still, the issue of information content, "the code" or "inscription" is inseparable part of understanding the control and regulatory patterns of cardiovascular system. Both aspects can be fruitfully combined in order to obtain fundamental understanding of cardiorespiratory integrative laws.