This Week in HRV — Episode 55

[00:00:00] Hey, welcome back to this Week in hrv. I'm Matt Bennett and this is the show where we dig into the most recent peer reviewed research on heart rate variability and autonomic health. Every episode grounded in actual published science and translated into something you can take somewhere useful. Before I get into today's lineup, let me say something about what a medical episode actually is in the context of this show, because I think it's worth being explicit. When I say medical, I don't mean this is going to be dense or clinical or only relevant if you're a healthcare provider. What I mean is that today's studies are all drawn from populations dealing with identifiable disease states or from research that is squarely aimed at answering clinical questions. How do we detect dysfunction earlier? How do we treat it more effectively? How does HRV fit into protocols we've been using for decades? I do this theme periodically because I think it serves a very specific purpose for our listeners. If you're someone who tracks HRV as a healthy person trying to optimize recovery or stress management, you might wonder why you should care about studies done in patients with lung disease or or rare autoimmune conditions. Here's the answer. Disease is not a separate world from health. It's further along the same continuum. The mechanisms of autonomic dysfunction that become unmistakable in severe disease. Chronic sympathetic overdrive, reduced vagal tone, blunted barrow reflex sensitivity. Those same processes are operating at lower levels in people who haven't been diagnosed with anything yet. Studying them at the extreme end is how we understand the system's full range. And the tools and protocols that are being validated in clinical populations are frequently the same ones that performance athletes and wellness practitioners end up adopting down the road. [00:01:24] Today we have seven studies. That's a lot for a single episode. And I'll say up front that every one of them deserves more airtime than I can give it. What I'll try to do is give you the core finding, the methodological context you need to evaluate it honestly, and the practical thread, what it means for how we think about HRV and what it might mean for how we use it. The range of territory we're covering today is striking. We start with something you encounter every day, noise. And we find out that your autonomic nervous system is keeping a very detailed log of your acoustic environment. We move from there to smoking and we finally get a measure of what it's actually doing to your heart rate in real life rather than in a clinical office. We look at a supplement called theacrin and whether it can buffer the HRV drop that follows a heart resistance training session. We go into the world of cardiac rehabilitation, specifically what happens to HRV when post open heart surgery patients combine structured exercise with resonance frequency breathing for eight weeks. We visit a rare and fascinating autoimmune disease called Morvan syndrome where HRV assessment is being used to map the extent of autonomic damage. We then look at a machine learning study asking whether your heartbeat alone, processed through a random forest classifier can tell you what stage of sleep you're in. And we close with a six month study following patients with chronic obstructive pulmonary disease through a physical activity coaching program where the most sensitive autonomic signal comes not from standard HRV metrics but from pointer a plot geometry, seven studies, seven angles, all of them pointing toward the same truth. The autonomic nervous system is the body's central integrating network, and HRV gives us one of the best windows we have into how it's functioning in real time. [00:02:53] One more thing before we start. I try to keep this show accessible across levels of HRV familiarity, and today's episode touches on some of the more technical Poincare analysis, Ewing's battery, and distributed lag models. I'll explain each as we go, but if any of the basics are still fuzzy, we what RMSSD measures, why the ratio of low frequency to high frequency power matters, how to interpret SDNN in different recording contexts. The show notes for this episode link to some prior episodes that build those foundations. You don't need them to follow today's content, but they are there if you want them. I'll also say a medical theme episode is a good moment to be explicit about what this show is and is not. This week in HRV is a science translation show. I am not a physician and nothing we discuss here is medical advice. What I'm doing is explaining what the published research says as accurately and as accessible as I can so so that you can bring more informed questions and frameworks to conversations with your own healthcare providers or to your own research if you work in this space. The studies we cover are real and the findings are as I describe them. But applying any of this to individual clinical decisions requires a clinical relationship that this show cannot and does not provide. The first study comes from the University of Michigan, published in the Journal of Exposure Science and Environmental Epidemiology in September 2026. The lead author is Shin Jiang, working with Sung Kyung Park, Lauren Smith, and Richard Neitzel. [00:04:07] The title is association between Environmental and Headphone Noise and Heart rate Observations from the Apple Hearing Study Cohort the Apple Hearing Study, if you haven't encountered it, is a massive real world epidemiological effort. Participants opted into research through the Apple Research app, shared data passively from their Apple watches and iPhones, and went about their normal lives while the devices continuously logged sound levels and physiological measurements. This is the kind of naturalistic data collection that was essentially impossible before consumer wearables, and it changes what kinds of scientific questions you can ask. [00:04:34] Instead of bringing people into a lab and playing sounds at them, you can observe thousands of real people in real environments and see what their bodies are actually doing. The researchers pulled a subset of consented adults from this cohort and organized them into four separate analytical groups defined by two dimensions. The first dimension was the sound source environmental noise, the ambient acoustic environment around you versus headphone sound, what you're personally piping into your ears. The second dimension was the exposure window. Short term defined as 20 minutes versus long term defined as 160 minutes. Within each of those four groups, they use stratified random sampling, pulling 20 participants from each combination of age, sex and ethnicity, giving them between 980 and 1,180 participants across the analyses. The HRV metric here is SDNN, the standard deviation of normal to normal intervals measured minute by minute from the Apple Watch. I want to note that this is 1 minute SDNN, which is shorter than the 5 minute or 24 hour windows typically used in clinical research and which has different interpretive standards but for the purpose of detecting real time autonomic responses to a time varying exposure, minute by minute is exactly what you want. The statistical model they used was a distributed lag nonlinear hierarchical Bayesian model. Let me unpack that phrase because it matters. Distributed lag means they were tracking how a noise exposure at one moment in time continued to affect HRV at successive intervals afterward, capturing the delayed physiological response that a simple correlation would miss. Nonlinear means they were not assuming that the relationship between noise and HRV is a straight line. They let the data shape the curve. Hierarchical Bayesian means They modeled individual variation explicitly rather than treating everyone as having the same response. This is a sophisticated and appropriate approach for the question they were asking now. The findings for every 10 decibel increase in environmental noise exposure, SDNN dropped by 6.8% in the short term 20 minute window and by 16% in the long term 160 minute window. Let me put the 10 decibel number in context. 10 decibels corresponds roughly to the difference between a a quiet library and a normal conversation, or between a conversational volume and a busy restaurant. We're not talking about standing next to a jet engine. This is the kind of acoustic fluctuation that happens in ordinary urban and suburban life multiple times per day. A 16% reduction in SDNN over a 160 minute exposure window is not trivial. SDNN is a broad marker of overall autonomic regulatory capacity, and a drop of that magnitude reflects meaningful autonomic stress that persists well beyond the moment of exposure. This fits the general stress model of noise. The idea that the nervous system perceives noise as a stressor activates a sympathetically biased arousal response and doesn't immediately recover when the noise stops. For headphone sound, the reductions were 7.1% short term and 7.4% long term. Notice that the two windows converge for headphones in a way they don't for environmental noise. One interpretation Headphone use tends to be episodic and self determined, so there's less cumulative dose difference between 20 and 160 minutes in the data than there is for environmental noise, which fluctuates unpredictably throughout the day. There was a lag specific finding that I found genuinely interesting. When they looked at moment by moment effects across time, they saw a brief initial increase in SDNN right at the onset of noise exposure, followed by a sustained decline. That initial uptick is consistent with an orienting response, a brief autonomic alert that the nervous system generates in response to any novel or unexpected stimulus. Your parasympathetic system briefly amps up as you pay attention. Then, if the noise continues and is perceived as a stressor, the sympathetic branch takes over and the vagal tone gets pulled down for a sustained period. The modifiers they found are also worth discussing because they tell us who's most vulnerable. Greater SDNN reductions from environmental noise were found in older adults, people with tinnitus, and people who rated their hearing as excellent. That last one seems counterintuitive. Why would excellent hearing make you more sensitive? The answer likely has to do with auditory attention. If you if you hear very well, you're processing more of the acoustic environment in detail, which translates to more cognitive and autonomic load. For headphone sound, the pattern was reversed. Worse hearing was associated with greater SDNN drops, probably because people with poorer hearing turn up the volume more, increasing the actual intensity of the signal entering the ear, even if they perceive it as normal. The practical implications are something I think about in the context of this show's audience. Most of us who track HRV are looking for the things we can modify. Sleep training, load, nutrition, stress management Environmental noise is not always on the list, but it probably should be. The commute with earbuds at high volume, the open plan office, the busy cafe where you're trying to work, the city apartment with street noise at night. Your autonomic nervous system is registering all of it, and the data suggests the cumulative effect is measurable. The authors are appropriately cautious about causal claims, and the Apple device user base is not demographically representative of the general population, but the scale of the study and the naturalistic design give these findings real weight. One practical implication I'd flag for HRV trackers Specifically, if you're trying to understand why a morning HRV reading is lower than expected, noise from the night before or during the night may be worth looking at as a variable. Sleep noise exposure has its own literature, and the pattern there is consistent with what this study found. In waking hours, traffic noise, intermittent sounds like garbage trucks or neighbors, or even running a fan in a way that introduces irregular sound variation can each suppress overnight autonomic recovery and in ways that translate to a depressed morning rmssd controlling for noise is not as glamorous as optimizing nutrition, timing or sleep stage distribution, but the evidence is suggesting it belongs in the same conversation. The second study is a systematic review and meta analysis published in Hypertension Research, a Nature Publishing group Journal, in September 2026. The senior author is Kazuomi Kario, one of the leading figures in ambulatory blood pressure research, and the first author is Yuya Akagi. The full author list is very long. This was a multi institution Japanese effort conducted as part of something called the Asakatsu BP Action Plan. The title is Differences in Ambulatory Blood Pressure and Heart Rate between Smoking and Abstinence Periods Among Current A Systematic Review and Meta Analysis Let me start with the measurement paradox that this study is directly addressing, because it's one of those situations in medicine where the standard of care produced a systematically misleading picture for a long time. When you measure a patient's blood pressure in a clinical office, standard protocols require them to rest quietly for several minutes before measurement. They're also typically advised not to smoke before the visit. [00:10:33] So what you're measuring in a smoker is their blood pressure in a state of temporary abstinence after a period of deliberate rest in an environment that is itself somewhat calming. This is about as far from their normal physiological state as you can get. What multiple epidemiological studies over the years have found is that smokers office Blood pressures tend to be similar to or even slightly lower than those of non smokers. This seemed to contradict everything we know about smoking's cardiovascular effects. Smoking acutely elevates blood pressure and heart rate through sympathetic nervous system activation. No epinephrine and epinephrine surge. The heart beats faster and harder. How could smokers have similar office blood pressure to non smokers? Ambulatory blood pressure monitoring resolves the paradox. Instead of measuring at one moment under artificial conditions, it samples blood pressure repeatedly across 24 hours while the person lives their normal life, including while they're smoking. And this meta analysis looks specifically at studies that compared ambulatory blood pressure within the same individuals across periods of active smoking versus periods of abstinence. This within person design is critical. It eliminates the confounding that plagues between person comparisons where smokers and non smokers differ in dozens of ways beyond just smoking status. The meta analysis identified four non randomized comparative studies that met criteria. Total sample size was 104 participants. The abstinence periods in those studies ranged from one day to one week. Using a random effects model, they estimated mean differences smoking period minus abstinence period for systolic blood pressure, diastolic blood pressure, and heart rate across different time windows. For systolic blood pressure, there was no statistically significant difference across any of the windows 24 hour daytime or nighttime. This is consistent with what office measurements have been showing for decades. Systolic is where the measurement paradox has always been most visible. For diastolic blood pressure, however, the picture changes. 24 hour diastolic was 2.67 millimeters of mercury higher during smoking periods with a confidence interval of 0.72 to 4.62. That confidence interval clearly excludes zero, meaning this is a real difference. Daytime diastolic was 3.17 millimeters of mercury higher during smoking periods. Nighttime diastolic did not reach significance and for heart rate the signal is the clearest of all. 24 hour heart rate was 6.47 beats per minute higher during smoking periods. The confidence interval was 3.98 to 8.96. Daytime heart rate was 7.19 beats per minute higher. Nighttime heart rate was 3.58 beats per minute higher and even that nighttime difference was statistically significant. Think about what a persistently elevated heart rate of nearly 7 beats per minute means for HRV. Heart rate and HRV are not in a simple inverse relationship, but they're related in important ways. Chronically elevated heart rate reflects reduced vagal tone. The parasympathetic branch of the autonomic nervous system is being suppressed throughout the 24 hour period. [00:13:08] Previous research cited in this paper, including a 1999 study by Manami and colleagues specifically looking at blood pressure and heart rate variability before and after smoking cessation, found that HRV improved after cessation. This meta analysis provides the ambulatory heart rate context that explains why during active smoking the sympathetic nervous system is running hot around the clock, not just in the moments after a cigarette. There's also the concept of masked hypertension to discuss here because the authors frame their work in that context. Masked hypertension is defined as normal office blood pressure but elevated blood pressure in daily life. It carries cardiovascular risk comparable to sustained hypertension, and smokers are disproportionately represented in this category. The diastolic elevation and the heart rate elevation captured in ambulatory monitoring are exactly the kind of signal that uncovers masked hypertension, the signal that office measurement systematically misses. The limitations here are real. Four studies and 104 participants is a small evidence base. The designs are non randomized abstinence periods varied across studies, but the within person methodology is methodologically strong and the consistency of direction across the included studies gives the findings credibility. [00:14:07] For practitioners and patients working with hrv, the implication is direct. If you're smoking, the baseline you're measuring against is not your true autonomic baseline. It's a baseline with chronic sympathetic arousal layered on top of it. And if you're helping someone who smokes track their HRV as part of a recovery or health monitoring protocol, this context matters. For how you interpret the numbers, I'll add something that isn't in this paper but follows naturally from it. The field of cessation research has a significant HRV component that doesn't get discussed much in mainstream health media. When smokers quit, HRV typically improves, sometimes substantially, over the weeks and months that follow. The previously cited Menami 1999 data showed HRV normalization post cessation. More recent longitudinal data is consistent with that. This means HRV tracking could theoretically serve a motivational function in cessation programs, providing real biological feedback that the autonomic system is recovering, which is the kind of tangible progress marker that behavioral change programs have consistently found to be motivating. That's a research area I think is worth watching over the next few years. [00:15:05] This next study shifts us into the supplementation research space, which I know is always a bit of a mixed bag. There's a lot of bad science in this area, but this one is methodologically careful and the question it's asking is genuinely interesting. It comes from the Journal of Vascular Diseases, published by MDPI with later author Gustavo Vieira de Oliveira. The full title is Effects of Theacrine Supplementation on Heart Rate Variability following a Resistance Exercise Session, A randomized double blind placebo controlled crossover pilot study. Theacrine, commonly sold under the brand name T Crene, is a naturally occurring alkaloid found in Camellia cucha, a Chinese tea plant, as well as in some coffee varieties. Structurally, it's related to caffeine and shares some of its mechanisms. It blocks adenosine receptors, which is the primary mechanism by which caffeine produces its stimulant effects and and it also appears to influence dopaminergic pathways. The difference from caffeine that tends to be cited in the literature is that tolerance may develop more slowly with the Akron, though the evidence on this is not definitive. It's found in pre workout formulas and focus supplements and it's been studied for effects on mood, energy and exercise performance, but its effects on post exercise autonomic recovery had not been examined before this study. The design is a crossover trial which is the appropriate approach when you're studying a within person question about a reversible intervention. [00:16:19] Seventeen young healthy resistance trained participants, eight female, nine male, completed two separate resistance exercise sessions separated by a washout period. In one session they took the accred 90 minutes before exercising. In the other they took a visually identical placebo. Neither the participants nor the testers knew which was which. At any given time. HRV was measured at three time immediately before supplementation, again 90 minutes after supplementation and just before exercise, and then 30 minutes after the exercise session concluded. The HRV metrics were RMSSD and hfnu, both of which reflect parasympathetic or vagal modulation. Here's what the data showed. In the theacrine condition, rmssd at the 90 minute post supplementation measurement was significantly higher than at the baseline pre supplementation measurement p value of 0.01. So 90 minutes after taking theacrine vagal tone was measurably elevated relative to where it started. Then after exercise, RMSSD in the theacrine condition came back down toward the pre supplementation baseline but the reduction was not statistically significant compared to either the pre supplementation or pre exercise readings. In the placebo condition the pattern was different. RMSSD dropped significantly after exercise compared to both the pre supplementation and pre exercise readings. That post exercise suppression of vagal tone is the expected and well documented response to a meaningful bout of resistance training. HFNU showed similar patterns to RMSSD across both conditions. The interpretation the Akron appeared to elevate vagal tone before exercise and that elevation may have provided enough of a buffer that the post exercise dip, while it happened, did not reach the same depth as it did under placebo. The absolute post exercise levels in the two conditions are not reported in the abstract, but the relative movement tells a story. The critical limitation is the time by condition interaction test. This is the formal statistical test you run to determine whether the two curves the acronym placebo are genuinely different trajectories rather than just two instances of normal variation. That test did not reach a statistical significance. The p value for the interaction was above the conventional 0.05 threshold. This means the study as a pilot trial with 17 participants cannot rule out the possibility that the pattern it found was due to chance. The effect on blood pressure, diastolic blood pressure and heart rate did not differ between conditions, which is reassuring. The Akron was not driving cardiovascular stress. I want to be precise about what this means and what it doesn't mean. A non significant interaction in a pilot study does not mean there's no effect. It means the study was not large enough to detect the effect. If it exists, the biological story is plausible. The akrine's action on adenosine receptors and dopaminergic pathways can plausibly support parasympathetic tone in ways that buffer the acute autonomic stress of intense exercise. The direction of the findings is consistent with that hypothesis, but this is a hypothesis that needs larger trials to confirm or refute. What this study contributes is a signal worth following. Post exercise HRV recovery is something a lot of practitioners and athletes care about, and the idea that a compound in the caffeine alkaloid family might protect rather than suppress vagal tone during the recovery window is genuinely novel. Whether theacrine is actually doing that or whether the effect size is meaningful enough to matter in practical terms requires more research. The broader question of resistance training and HRV is worth a brief comment here because the pattern this study is testing against is well established and clinically meaningful. Heavy resistance training acutely suppresses vagal tone, RMSSD falls sympathetic markers rise. In healthy athletes, this suppression reverses over 24 to 48 hours and tracking the pace of that recovery is one of the ways HRV monitoring provides practical value for training load management. But in clinical populations, cardiac patients, people with metabolic disease, those on certain medications, the post exercise HRV diploma can be deeper and longer lasting and understanding what modulates it has real management implications. The ocarina is one candidate, there are others omega 3 supplementation, certain anti inflammatory strategies, sleep quality might all of which have preliminary data suggesting they influence the pace of post exercise autonomic recovery. It's a rich area that I hope to come back to in future episodes with more data in hand. This one is from Kuraeus, authored by Puvishnu T Devi and the title is Effective the Otago Exercise Program and Resonance frequency breathing in patients following coronary artery bypass grafting Coronary artery bypass grafting is one of the most common major cardiac surgeries performed worldwide. The procedure involves taking blood vessels from elsewhere in the body, typically the leg or internal mammary artery, and grafting them to bypass blocked coronary arteries, restoring blood flow to the heart muscle. It's effective at relieving symptoms and improving survival in the right patient populations. But the surgery itself the cardiopulmonary bypass, the anesthesia, the physiological stress causes profound disruption to autonomic function that can persist long into the recovery period. Post CABG patients have consistently documented HRV depression. Their resting heart rate is elevated, heart rate recovery after exertion is slower, the parasympathetic nervous system is suppressed while sympathetic tone dominates, and this autonomic imbalance is independently associated with worse outcomes including arrhythmia risk and mortality. Cardiac rehabilitation exists precisely to address this and the research question this study is asking is whether adding resonance frequency breathing to a standard exercise rehabilitation protocol produces measurable improvements in hrv. The Otago Exercise Program for Context is a structured intervention developed in New Zealand originally designed to reduce falls in older adults through a combination of leg strengthening exercises and balance training. It was delivered three times per week. In this study. The resonance frequency breathing component was 15 to 20 minutes of paced breathing at six breaths per minute done daily. Six breaths per minute is resonance frequency for most adults. The scientific basis for this is the coupling of respiration and heart rate through respiratory sinus arrhythmia, the natural speeding up of the heart during inhalation and slowing during exhalation that is driven by the vagus nerve when breathing rate slows to around six cycles per minute. The respiratory cycle synchronizes with the body's natural oscillation and blood pressure regulation driven by the baroreflex. This synchronization amplifies heart rate oscillations dramatically and with repeated practice, strengthens baroreflex sensitivity and increases resting hrv, the evidence base for resonance frequency breathing as a deliberate autonomic conditioning tool is among the strongest in the field. 40 patients between 40 and 60 years of age who had undergone CABG participated in the eight week program. The HRV assessment was a five minute ECG recording with RMSSD as the primary metric. This is a time domain measure reflecting parasympathetic beat to beat variability highly appropriate for tracking short term vagal changes. Before the intervention, mean RMSSD was 18.03 milliseconds. After eight weeks it was 21.96 milliseconds. That's an increase of 3.93 milliseconds and it was statistically significant with P 0.001. For reference, post CABG patients typically show RMSSD values substantially below age matched healthy adults where 20 to 30 milliseconds might be a rough normal range for someone in their 50s under standard clinical HRV protocols. So even the post intervention value of 21.96 milliseconds is still modest but the direction and magnitude of change matter. Clinically, the physical performance measures improved significantly as well as on the timed up and go test where you stand from a chair, walk three meters, turn walk back and sit performance improved from a mean of 17.95 seconds at baseline to 14.68 seconds at eight weeks. The chair stand test measuring lower limb strength and functional capacity improved from 13.03 to 15.78 repetitions. Balance measures also moved in the right direction. I want to address the study design honestly because it matters for how much weight you put on these results as this was a single group pre test posttest design, there was no control group that received neither the exercise program nor the breathing protocol. That means we cannot distinguish improvement due to the intervention from improvement that would have happened anyway through natural post surgical recovery over eight weeks. We also cannot separate the effects of the Otago exercises from the effects of the resonance frequency breathing. They were delivered as an inseparable package and a factorial design with four groups. Exercise alone, breathing alone combined control would be required to answer that question. I also want to say something about the mechanism of resonance frequency breathing that I think helps explain why the effect sizes in this literature are as large as they are when you breathe at six cycles per minute, the inhalation phase drives a heart rate increase and the exhalation phase drives a heart rate decrease. This is respiratory sinus arrhythmia at resonance frequency. The duration of each phase aligns precisely with the body's intrinsic blood pressure oscillation cycle which runs at approximately 0.1 Hz. The respiratory driven heart rate swings reinforce These blood pressure oscillations and the combined signal strongly activates the baroreflex, the pressure sensitive feedback system in the carotid arteries and aorta that regulates blood pressure and heart rate moment to moment. With repeated daily practice, baroreflex sensitivity increases. The system becomes better at detecting and correcting small deviations, which shows up as larger amplitude HRV oscillations at the resonance frequency typically captured in the LF band of frequency domain. HRV analysis this is not just relaxation, it is targeted training of the autonomic reflex arc and the eight week timeline in this study aligns with what other baroreflex training research has shown as the period needed to produce durable structural adaptation. These are real limitations and I want to be transparent about them. What this study does within those constraints is establish feasibility. This combination can be delivered to post CABG patients. They tolerate it and the outcome measures move in the right direction by a meaningful amount. For clinicians designing rehabilitation programs, this is useful practical evidence even in the absence of a randomized controlled trial, particularly given the low risk and low cost of the intervention. The resonance frequency breathing component in particular stands out to me as undertapped in cardiac rehabilitation. The mechanistic evidence is strong, it requires minimal equipment or training and the autonomic benefits extend beyond just HRV to include blood pressure regulation and reduced anxiety. If you're a practitioner working with post cardiac surgery patients, this study adds to a growing body of work suggesting that paced breathing at resonance frequency belongs in the protocol. Here's the study that surprised me most when I was putting this episode together. It comes from the Revue Neurologique, a French neurology journal published by Elsevier, authored by C. Nandan G. And Baraj K. Arjun, K. Udupa, K. Prasad and colleagues. The title is Cardiovascular Autonomic Dysfunction in Morvan Syndrome. A Comparative study of heart Rate variability and cardiovascular autonomic reflex tests Most people have never heard of Morvan syndrome. It's a rare autoimmune encephalopathy. Let me break down those words. [00:26:24] Autoimmune means the immune system is attacking the body's own tissues. Encephalopathy means there's brain dysfunction. Morvan syndrome specifically involves antibodies targeting proteins in the voltage gated potassium channel complex, most commonly CASPR2 and LGI1. These proteins help regulate the electrical excitability of neurons. When antibodies block or destroy them, neurons fire too readily, producing a distinctive constellation of symptoms. The clinical triad of Morvan syndrome has three components. First, neuromyotonia, continuous spontaneous discharge from peripheral motor nerve terminals that causes visible muscle rippling under the skin, cramps, muscle stiffness, and what's sometimes described as a kind of buzzing or electrical sensation. Second, encephalopathy, confusion, memory impairment, personality change, agitation, hallucinations, and characteristically severe insomnia. Some patients go weeks with virtually no sleep. Third, and this is the component that brings us into HRV territory Dysautonomia Widespread disruption of the autonomic nervous system the autonomic symptoms in Morvan syndrome can range from uncomfortable to life threatening. Excessive sweating and hypersalivation are common. Urinary incontinence, blood pressure instability, swings between hypertension and hypotension, tachycardia and in severe cases, cardiac arrhythmias. The autonomic involvement in this disease reflects the fact that the same ion channel proteins that are being targeted by the antibodies are expressed not just in the brain and peripheral nerves but in the autonomic ganglia, the relay stations through which the autonomic nervous system controls organs. The research question this group is asking is how systematically are we detecting and characterizing this autonomic dysfunction? [00:27:53] And can a combination of HRV analysis and the Ewing's Battery of cardiovascular autonomic reflex tests give us a more complete picture of the damage? Ewing's Battery for listeners who haven't encountered it, was developed by David Ewing and colleagues in Edinburgh in the 1970s and remains the clinical gold standard for comprehensive autonomic assessment. It's a structured set of tests that probes specific arms of the autonomic reflex arc. The heart rate response to deep breathing evaluates vagal function at the sinus node. The heart rate responds to the Valsalva maneuver bearing down against a closed airway tests both sympathetic and parasympathetic responses. The heart rate response to standing from lying down evaluates the initial sympathetic acceleration followed by vagal slowing that occurs with postural change, blood pressure responses to sustained hand grip and orthostatic challenge. Evaluate sympathetic vasomotor function. Each test has established normative thresholds and a scoring system that grades autonomic function from normal to severely impaired. The complementarity of HRV and Ewing's battery is important. HRV gives you the resting state. It tells you what the autonomic nervous system's baseline activity looks like when the body is not being challenged. Ewing's tests give you the dynamic response. They stress the system with known provocations and measure how well the reflexes respond. Together they tell you both the baseline and the reserve capacity. In a disease like Morivan syndrome, where the immune attack may be damaging the system at multiple levels peripheral nerve, autonomic ganglia, central control. You need both perspectives to understand the full extent of the dysfunction. The clinical implication for HRV monitoring and Morvan syndrome is about early detection and tracking. The disease has a variable course. Some patients respond well to immunotherapy, intravenous immunoglobulin, plasma exchange, steroids and go into remission. Others have a more aggressive trajectory. If comprehensive HRV assessment can identify autonomic deterioration before it becomes clinically obvious or contract treatment response more sensitively than symptom scores, that's a meaningful addition to clinical management. There's also a diagnostic utility argument here. Morvan syndrome is frequently misdiagnosed for months or years before the correct antibody panel is run. One reason is that the autonomic symptoms sweating, tachycardia, blood pressure instability are individually non specific when they appear alongside HRV patterns consistent with severe autonomic dysfunction and the specific Ewing's battery profile associated with peripheral ganglionic damage. That combination has diagnostic specificity that individual symptoms do not. In diseases where early diagnosis enables earlier treatment and better outcomes, tools that sharpen the diagnostic picture matters. The broader lesson, and this is the one I want listeners to take from this study, even if they'll never personally encounter a patient with Morvan syndrome, is that the autonomic nervous system is a primary target in many neurological and autoimmune conditions. Parkinson's disease, multiple system atrophy, diabetic neuropathy, autoimmune encephalopathies. All of these damage the autonomic reflex arc, and in many of them autonomic symptoms precede more visible motor or cognitive symptoms by years. [00:30:33] HRV has the sensitivity to detect changes in autonomic tone that precede the clinical threshold. Studies like this one in rare diseases are building the evidence base for what HRV profiles look like in different pathological conditions, and that knowledge ultimately benefits the broader field. This study comes from the National Medical Journal of India, published in June 2026 with authors Suvaradib Chakraborty, Manish Goyal, Paritosh Goyal and Priyadarshani Mishra from the All India Institute of Medical Sciences in Bhubaneswar. The title is Artificial Intelligence Based Automated Sleep Staging using heart Rate Assessment of Performance and clinical Prospects. [00:31:09] Sleep staging is the process of classifying what stage of sleep a person is in at any given moment during the night wake, light sleep, moderate sleep, deep slow wave sleep or REM sleep. These stages have profoundly different physiological characteristics and the proportion of time spent in each one, as well as the sequencing of stages across the night, is clinically meaningful. Poor sleep architecture Too little Anthria rem, frequent transitions Prolonged wakefulness after sleep onset is associated with cardiovascular disease, metabolic dysfunction, cognitive impairment, and reduced immune function. The gold standard for sleep staging is polysomnography, a comprehensive overnight recording in a sleep laboratory involving eeg electrodes on the scalp to measure brain waves, electrodes on the face to track eye movements and muscle activity, respiratory sensors, ECG, and pulse oximetry. A trained sleep technologist manually scores the recordings using standardized criteria. It's accurate. It's also expensive, inconvenient, and only captures a single night under artificial conditions that may not reflect the patient's typical sleep. The question this group is asking is whether you can do a meaningful job of sleep staging using only the heartbeat, specifically using HRV features extracted from ECG data. If you can, then sleep staging becomes potentially accessible through any wearable device that records RR intervals and whether that's a chest strap, a smartwatch, or a clinical cardiac monitor. This would enable longitudinal sleep monitoring at scale in populations where polysomnography is impractical. The data set they used is from the PhysioNet Computing and Cardiology Challenge 2018, a publicly available database of overnight polysomnography recordings. They selected 645 subjects, extracted ECG data, and applied linear interpolation to correct for ectopic beats and outliers from the corrected RR interval series. They computed HRV features across all three domains. Time domain metrics like SDNN and rmssd frequency domain metrics capturing low frequency and high frequency, power and nonlinear metrics. All three domains carry information that's relevant to sleep stage classification. Time domain metrics reflect overall variability. Frequency domain reflects the oscillatory structure of autonomic control. Nonlinear metrics capture complexity and dynamics that aren't captured by the other two. They train two different models on these features. The first is a random forest classifier, an ensemble method that builds many independent decision trees during training and aggregates their predictions. Random forests are robust to overfitting, handle nonlinear relationships well, and produce interpretable feature importance rankings. The second model is a bidirectional long short term memory network, a type of recurrent neural network that processes sequences in both forward and backward time directions simultaneously. BLSTMs are powerful for sequence modeling tasks where context in both directions matters and and sleep staging has obvious temporal structure. What stage you're in now depends partly on what stage you were in five minutes ago and what stage you'll be in five minutes from now. The random forest classifier achieved a validation accuracy of 79.6% across five sleep stages, the BLSTM achieved 74.7%. The Random Forest outperformed the deep learning model, which is not unusual in healthcare applications where data sets are relatively small and the signal is structured enough that simple ensemble methods can capture it efficiently. One of the methodologically important contributions here is the inclusion of the sleep epoch index as a feature. Every sleep epoch, each 32nd window of the recording has a position in time within the night. The researchers included that positional information as an input variable, and they found it improved classification accuracy meaningfully. The intuition is correct Sleep architecture follows a biological rhythm in three Deep sleep tends to dominate the first third of the night. REM periods lengthen across the night, with the longest REM episode often occurring in the final third. [00:34:41] If your model knows where in the night a given epoch falls, it can use that knowledge alongside the HRV features to make a better guess about the stage. This is an important methodological lesson for anyone building or evaluating HRV based classifiers. Temporal context often matters as much as the measurement value itself. A model that ignores the sequential structure of sleep staging is throwing away useful information. The external validation on the Hyglan and MITISCH Centrum dataset 43 subjects from a completely separate data collection produced an accuracy of 78.9%, a Cohen's cap of 0.70, and a macro F1 score of 0.789. Cohen's cap of 0.70 is generally considered to represent substantial agreement. The macro F1 score means the model is performing reasonably well across all five classes, not just the majority one, which matters because N1 and Wake are underrepresented in overnight recordings and a model that ignores them to boost accuracy would be clinically useless. The limitations are honest ones. Five stage accuracy in the upper 70s leaves real errors, particularly in distinguishing the lighter stages from each other and from brief wakefulness stages. N1 and wake share similar HRV characteristics because the body is in a similar autonomic state in both. Getting those right requires the brainwave information that ECG alone cannot provide. This is a screening and monitoring tool, not a replacement for clinical polysomnography in situations where accurate staging is required, but for the clinical context where this matters most. Detecting sleep disordered breathing in cardiac patients Tracking sleep quality longitudinally in chronic disease populations Population level research where Polysomnography is cost prohibitive 79% accuracy from a device that's already being worn is a major practical advance. The infrastructure for widespread deployment of this kind of classification already exists in current generation wearables. What this study contributes is the validated algorithmic approach that makes it clinically credible. I also want to connect this study to the consumer HRV space because I think the implications are closer to home than they might appear. If you wear a chest strap or a ring to bed and it gives you a sleep score, that score is almost certainly based on some version of what this paper is testing. HRV features possibly combined with movement data used to classify sleep stages. The accuracy numbers in this study give you a realistic benchmark for what that estimate is likely to be. Around 79% at the five stage classification level sounds good, but it means roughly one in five epochs is misclassified For a healthy sleeper with typical architecture. That error is probably distributed in ways that don't distort the summary statistics much, but so for a person with atypical sleep patterns, irregular staging, frequent brief awakenings, sleep apnea those errors can compound into a meaningfully inaccurate picture. This doesn't mean you should distrust your sleep data entirely. It means understanding what it can and cannot tell you precisely, which is something every user of these technologies should have the context to do. The final study today is the one I've been building toward and I want to give it the time it deserves. This comes from PLOS One, published in August 2026 by Fian Hermans and Helene Dubmeyer at Ghent University Hospital, along with a large multidisciplinary team including pulmonologists, physiotherapists and respiratory scientists. The title is the Relation between Physical Activity and Cardiac Autonomic Dysfunction in Patients with Chronic Obstructive Pulmonary Disease. Let me start with the population context because it shapes everything that follows copd. Chronic obstructive pulmonary disease is a progressive, largely irreversible condition and characterized by persistent airflow obstruction typically caused by long term exposure to cigarette smoke or other inhaled irritants. It's one of the leading causes of death worldwide. We tend to think of it primarily as a lung disease. The airways become inflamed and narrowed, the alveoli are destroyed, gas exchange is compromised, but its effects are systemic in ways that are increasingly recognized. Patients with COPD have significantly lower HRV than healthy age matched adults. Their resting heart rates are higher. Heart rate recovery after exercise is slower and less complete. All three of these markers reduced hrv, elevated resting heart rate, impaired heart rate recovery are independently associated with worse outcomes in copd, including increased risk of acute exacerbations and higher mortality. The autonomic dysfunction in COPD is not a side effect of the disease it's part of the disease process driven by chronic hypoxemia, systemic inflammation, and the structural changes in the lung that affect the vagus nerve. Physical activity is severely limited in copd. Patients move less than healthy peers and and that reduced activity has its own downstream effects on autonomic function, a vicious cycle where the disease limits exercise, the reduced exercise worsens autonomic tone, and the worsened autonomic tone contributes to disease progression. The question this study is investigating is whether we can detect a meaningful signal of autonomic improvement in patients who increase their physical activity and which HRV metrics are most sensitive to that signal. The study was nested within a larger randomized controlled trial of physical activity coaching in COPD. 36 patients at Ghent University Hospital were enrolled in the mean age was 69 years, 69% male. They ranged from mild to severe COPD 56% had mild to moderate disease, importantly 28%. Ten patients were on stable doses of beta blockers, which suppress heart rate and HRV directly. The researchers ran separate analyses for the full group and for the subset not on beta blockers throughout. The autonomic assessment approach here is methodologically impressive rather than a brief clinical recording, patients Wore the Polar H10 chest sensor with for four consecutive nights at baseline and again at six months. Follow up. Heart rate and RR intervals were continuously logged during sleep. The most standardized and least movement contaminated window for HRV measurement data was processed through Kubio's HRV Scientific software and the analysis included four distinct HRV metrics SDNN and RMSSD in the time domain, LF HF ratio in the frequency domain, an SD2 SD1 ratio from Poincare a plot analysis and in the nonlinear domain. Let me explain the Poincare plot analysis in enough detail that the finding makes sense. A Poincare plot is a scatter plot where each point represents a pair of consecutive RR intervals, the current interval on the x axis and the next interval on the Y axis. When you plot thousands of these pairs from an overnight recording, they form a characteristic elliptical cloud. The short axis of that ellipse called SD1 represents the beat to beat variability that is driven primarily by vagal activity. The long axis you called SD2 represents the longer term variability that reflects the overall modulation of heart rate over time by both branches of the autonomic nervous system. The ratio SD2 over SD1 is a measure of sympathovagal balance. A high SD2 over SD1 ratio means the ellipse is elongated along its long axis relative to its short axis. This pattern is associated with sympathetic dominance. A lower ratio where the ellipse becomes rounder reflects a more parasympathetic vagal tone. When we say a study is pointer a heavy what we mean is that this nonlinear geometric analysis is not just one metric among many it's a central tool for interpreting the autonomic state of the system. The physical activity Intervention was a six month coaching program delivered via a Fitbit Charge 4 wearable, a smartphone app and periodic coach contact. Patients were randomized to either a full coaching program with individualized adaptive step goals and regular feedback or a lighter version with a fixed step goal. Both groups were pooled for the analyses. In this sub study, the result of the coaching on physical activity was modest. Median change in daily step count was 525 steps per day, considerably less than the target and not statistically significant. Movement intensity during walking, measuring how vigorously people moved when they did walk also did not change significantly at the group level. Similarly, the group level changes in HRV metrics between baseline and six months were not statistic statistically significant. On the surface this might look like a null result, but the correlation analyses are where the study gets interesting and this is where the Poincare geometry earns its place. When the researchers looked at the relationship between changes in physical activity and changes in HRV over the six months that is among the individuals who actually moved more did their autonomic profiles improve. They found meaningful associations, particularly in the nonlinear domain. Change in movement intensity during walking was inversely associated with change in SD2 SD1 ratio with a Spearman coefficient of minus 0.37 in the full sample exclude the beta blocker users and that coefficient moves to minus 0.51. In a study of 36 patients with a noisy, complicated disease process that's a moderate and clinically meaningful association. Patients who walk More vigorously over six months showed a decrease in their SD2 SD1 ratio, a shift toward greater vagal dominance, a more parasympathetic autonomic balance in the beta blocker excluded. Subgroup increase in daily step count was also inversely associated with change in resting heart rate RS -0.48, change in LF HF ratio RS -0.35 and change in SD2SD1 ratio RS -0.28. All of these point in the same direction. More physical activity was associated with improved autonomic balance, lower resting heart rate, less sympathetic dominance in the frequency domain and a more rounded peinchere ellipse. The discussion section makes a point I want to highlight explicitly. The authors note that the improvements in sympathovagal balance were more pronounced in the nonlinear measures SD2 SD1 than in the linear time domain or frequency domain measures. They cite the broader literature suggesting that nonlinear HRV indices are often more sensitive to subtle autonomic changes than SDNN or rmssd. This aligns with what we've seen in other contexts. Point rate plot geometry encodes information about the dynamic structure of heart rate variability that a single statistical summary number cannot capture. What does this mean for how we think about physical activity in chronic disease? The study is careful about causation. With no control group and small sample size, the correlational findings can't be turned into a definitive clinical recommendation. But the direction of the signal and the fact that it's coming out of the nonlinear analysis when the linear analysis was quieter is telling us something important. The quality and intensity of physical activity, not just the quantity, is what shifts the autonomic balance. Walking more steps at the same pace did not produce as clear a signal as walking more intensely. This has direct implications for how we design activity interventions in COPD and and likely in other chronic disease populations. Not just more steps, but better steps the authors also make a strong argument for routine HRV monitoring in COPD clinical care. The Polar H10 they used is widely available and relatively inexpensive. Kubios HRV standard software has a free version. The measurement window four nights of passive recording is demanding but feasible for patients who need only to wear a chest strap while sleeping. The information it yields resting autonomic tone puncture, a geometry baroreflex sensitive frequency components is richer and more dynamically relevant than the static spirometry measures that currently dominate COPD assessment. This study adds to the case that HRV belongs in routine COPD management. I want to spend a moment on the walking intensity finding because I think it has implications beyond copd. The standard paradigm in chronic disease activity research has been step counting. Get patients to move more, measure the output in steps per day. This study is suggesting that how you move those steps matters as much as whether you take them. The Spearman coefficient for walking intensity versus SD2SD1 change was stronger than the coefficient for step count versus SD2SD1 change, particularly after removing the beta blocker confounding. This is consistent with the exercise physiology literature on cardiovascular adaptation. More broadly, it's intensity that drives adaptation. Whether we're talking about cardiac output, oxygen uptake or or apparently autonomic balance, walking faster, even a little, appears to be a more powerful autonomic signal than walking more at the same pace. If that holds across replications and in other chronic disease populations, it could reshape how activity coaching programs are designed not just walk more, but walk with some purpose and speed. That's a simple message, and it might be a very powerful one. We've covered a lot of ground today, and I want to take a few minutes to pull the threads together because I think there are themes here that are worth naming explicitly. [00:45:58] The first theme is measurement context. Four of today's seven studies made a point of measuring autonomic function in natural real world settings rather than under artificial clinical conditions. The Apple Hearing study used ambient wearable data. The Smoking Meta analysis used ambulatory monitoring rather than office readings. The COPD study used four nights of sleep time recording. The AI Sleep Staging study used a publicly available overnight polysomnography database. All of these approaches are capturing something more ecologically valid and than what you get in five minutes of rest in a clinical office. And the scientific community is increasingly recognizing that ecological validity matters for understanding how HRV actually varies and what it actually predicts. The second theme is the sensitivity of nonlinear metrics. This came up most explicitly in the copd study where SD2SD1 from pointer a plot analysis showed associations that the conventional time domain metrics missed. It also came up implicitly in the AI sleep staging study where nonlinear features were part of the input set that drove the classifier's accuracy. If you're evaluating HRV software or wearables and trying to decide what metrics matter, this body of evidence consistently argues for going beyond RMSSD and SDNN to include Poincare derived and other nonlinear measures, particularly when you're tracking subtle changes in complex systems. The third theme is the connection between respiration and the autonomic nervous system. The CABG study is the most direct example. Resonance frequency breathing at six breaths per minute deliberately exploits the coupling between breath and heartbeat to train the baroreflex and increase rmssd. But this mechanism is relevant across multiple studies. The COPD study is measuring autonomic function in patients whose respiratory physiology is fundamentally compromised. And the two things, like lung disease and autonomic dysfunctions are not separate problems they share mechanisms and interact. Understanding the respiratory autonomic interface is foundational to understanding what HRV means in any population. And the fourth theme is clinical reach from a rare autoimmune encephalopathy that affects perhaps a few hundred patients worldwide to A disease affecting hundreds of millions globally. Today's studies span an enormous range of clinical contexts. What they share is the use of HRV as a window into autonomic function. A window that provides information that other measures miss, that can be captured non invasively and that tracks meaningful physiological change over time. The evidence base for HRV's clinical utility keeps growing and the breadth of its application keeps expanding. There's also a fifth thread I want to name, one that cuts across all seven studies, even though it's not explicitly stated in any of them. The Democratization of Measurement the Apple Watch logs SDNN continuously in real people going about their lives. The Polar H10 chest strap records four nights of HRV in COPD patients at home. The PhysioNet dataset enables any research group in the world to train sleep staging classifiers without running a single polysomnography study themselves. The Kubio software has a free tier accessible to any clinician willing to learn it. Ten years ago, most of what was done in today's studies required either a research lab or a clinical sleep center. The barriers are lower now, and they're going to keep falling. What that means for the pace of discovery and for who gets to participate in that discovery is something I find genuinely exciting. The research is moving faster because the infrastructure is moving faster, and episodes like this one are part of what I hope is a broader project of making sure that acceleration translates into understanding. That's the episode I hope you came away from today's show with a richer picture of what HRV can tell us about how the body is doing in disease and recovery. In the everyday acoustic environment you move through and in the quiet hours of sleep when when your nervous system is doing some of its most important work. One thing I keep returning to when I put together an episode like this one, seven studies across seven very different contexts, is how consistent the underlying story is. The autonomic nervous system does not segment neatly into clinical and non clinical, healthy and unhealthy, normal and abnormal. It's a continuous system under continuous influence from continuous inputs. The air you breathe and how well your lungs process it, the sounds arriving at your ears, the cigarette in your hand or not in your hand, the weight you lifted this morning, the surgery you had last month, the immune cells circulating in your bloodstream, the dreams you had last night. HRV is not a perfect readout of any of these individual influences, but is a summary measure of where the system sits in the space between sympathetic and parasympathetic dominance. It captures something that nothing else captures as cleanly, as continuously and as non invasively. The studies we discussed today push the field forward in different ways. The Apple Hearing study gives us ecological validity at scale. The Smoking meta analysis corrects a long standing measurement artifact that was generating misleading clinical conclusions. The Theacrin trial opens a new research question about supplementation and post exercise autonomic recovery. The CABG Rehabilitation study provides feasibility and directional evidence for a low cost intervention that deserves a proper randomized trial. The Morvan Syndrome study applies HRV in a diagnostic and monitoring context where it could meaningfully change clinical outcomes. The AI Sleep Staging study provides a validated framework for clinical grade sleep assessment from consumer accessible hardware and the COPD study demonstrates that the most sensitive marker of autonomic change in a challenging population may be found not in the conventional metrics, but in the geometric structure of the Poinchara Plot if this episode was useful to you, please share it with someone who would find it valuable. If you work in clinical practice or research and you want to talk about any of these studies in more depth, reach out. I read everything and I try to respond and I'll be back next week with more research, more translation and more time with the data that's changing us how we understand the autonomic nervous system. Until then, keep tracking, keep moving and take good care of your nervous system. It is the only one you have got and science is making it increasingly and undeniably clear that it is worth paying very close attention to.