This Week in HRV - Episode 58

[00:00:00] Welcome to this Week in hrv, the podcast that brings you the latest peer reviewed research on heart rate variability and what it means for your health, your performance and your recovery. I am your host Matt Bennett, co founder of Optimal hrv, and if you are new here, the idea behind this show is simple. Every week we take the most interesting recent studies on HRV science and break them down in a way that is genuinely useful for whether you are a clinician, a coach, an athlete, a researcher, or just someone who wants to understand their own nervous system better. We try to be honest about what the research actually shows, which means we celebrate good findings and we are equally willing to say when something turns out to be more complicated than we hoped. That kind of honest engagement with science is, I think, the only way this field makes real progress and it is what you deserve As a listener. [00:00:57] Before we get into the research, I want to share our standard medical disclaimer. The information in this podcast is for educational and informational purposes only and is not intended to diagnose, treat, cure or prevent any medical condition. Please consult with a qualified healthcare provider before making any changes to your health regimen, training program or clinical practice. [00:01:22] Heart rate variability or HRV is a powerful and scientifically validated tool for understanding autonomic nervous system function, but it is one piece of a much larger picture and context always matters. Nothing I say here should be taken as personalized medical advice, and individual responses to interventions vary enormously. Always work with qualified professionals when making decisions about your health. [00:01:49] Today is an episode about performance, and I mean that in the broadest possible sense. We are not just talking about athletic performance, though one of our studies does involve competitive wrestlers in the midst of a serious training program. We are talking about performance as a human phenomenon that spans the golf course, the cognitive laboratory, the highway, and the wrestling mat. We are looking at what the nervous system does when we are trying to focus under pressure, when we are fighting mental exhaustion, or when we are dangerously drowsy behind the wheel, and when we are adapting or not adapting in the ways we expected to sustained physical stress. All four studies we are covering today use HRV as their central measurement tool, but they ask very different questions and approach the data from very different methodological angles, which is one of the things I find most exciting about this particular collection of papers. We start with a study on strategic self talk in golf pudding, which might sound like an unusually specific topic, but turns out to be a beautifully designed investigation into something how internal attentional strategies change the physiology of skilled performance and whether heart rate variability can serve as a window into that process. [00:03:04] From there, we move to a study that deploys an innovative sensor cushion, one that you sit on rather than wear, to try to capture HRV continuously during a sustained cognitive task under nap deprivation conditions, asking whether the nervous system leaves a detectable trace of mental fatigue in the cardiac autonomic signal. After that, we examine a paper about drowsy driving that asks whether machine learning applied to HRV features extracted from an electrocardiogram recording can reliably distinguish a fully rested driver from one who has been awake for approximately 30 hours and and we close with what I think is one of the most methodologically ambitious studies we have covered in a long time, a longitudinal investigation of competitive wrestlers that uses principal component analysis and hierarchical clustering to reveal that individual athletes respond to identical training in fundamentally different ways and that multivariate HRV analysis can characterize those differences in reproducible, meaningful terms. [00:04:13] Let us begin with the Self Talk Study. The full title is Exploring Attentional Mechanisms of Strategic Self Talk through Heart Rate Variability in a Golf Putting Task among Novices, and it comes from a team of eight researchers Eminoid Zoram Pazakis, Theodoros Proskinitopoulos, Orestes Pannulas, Evangelos Galanis, Evgenia Nicolakopoulo, Nikos Comautos, Yanis Theodorakis, and and Antonis Hatze Georgiadis. If you follow the sports psychology literature, Hatze Georgiadis is a name you will recognize. He has spent the better part of two decades studying how athletes use self talk to regulate their performance, and he has been central to building an empirical foundation for what was once considered a fairly soft intervention. This study represents a natural and sophisticated extension of that work, moving beyond the question of whether self talk improves performance to the more mechanistic question of how it does so, and specifically, whether the mechanism involves measurable changes in autonomic nervous system function. [00:05:18] The theoretical scaffold the researchers are working from is drawn from attentional theories of motor learning and expert performance. [00:05:25] When we are acquiring a new motor skill, whether that is a golf putt, a basketball free throw, or or a surgical suture, initial learning is effortful and explicitly controlled. We must consciously allocate attention to the components of the movement the position of our feet, the angle of the club face, the line of the putt, the pressure of the grip. [00:05:50] This conscious, effortful processing is cognitively demanding and keeps the nervous system in a relatively activated state. [00:05:58] As practice accumulates and the movement becomes more automatic, execution no longer requires the same degree of explicit attentional control. The movement proceeds more efficiently with less conscious oversight, and the nervous system can shift toward a less metabolically costly regulatory mode. [00:06:17] Strategic self talk is thought to work in part by directing the performer's attention in ways that align with this shift toward efficient automatic processing. It helps the athlete focus on what matters, suppress distracting thoughts, and execute with less conscious interference. [00:06:35] The key question this study asks is, can we see that attentional shift in the physiology of the autonomic nervous system as measured by HRV? To test this, the team recruited 40 male sports science University students with no prior experience in golf pudding. The choice of novices is deliberate and methodologically sound. In beginners, the trajectory from effortful conscious control toward more automatic processing happens rapidly across the early stages of learning, which means the effects of an intervention on that trajectory are large and detectable. With experienced performers, the baseline is already close to automatic and the room for improvement is smaller. Participants were randomly assigned to either a control group or or a self talk experimental group, and both groups followed an identical protocol across four a baseline assessment, two training sessions, and a final assessment. Both groups practiced the same putting drills for the same amount of time with the same technical instruction. The difference was that the experimental group additionally worked with a sports psychology specialist to develop personalized self talk cues in individually crafted task specific words or short phrases that each participant would use during putting to direct their attention in productive ways. These were not generic motivational phrases they were tailored to the attentional demands of the task and meaningful to the individual performer. The control group received no such attentional training at both the baseline and final assessment sessions. The researchers measured pudding performance, the primary behavioral outcome, and heart rate variability during the task using root mean square of successive differences, which we call RMSSD as the primary HRV metric. [00:08:24] RMSSD is the most widely used time domain measure of parasympathetic nervous system activity in sport and exercise science contexts. It reflects the B2B changes in heart rate that are driven by the vagus nerve and and higher values indicate greater parasympathetic influence on the heart. The researchers analyzed RMSSD not just as a single aggregate value but across different phases of the pudding task, asking whether the temporal pattern of autonomic activation differed between groups and changed across the training period. [00:08:58] The performance results were clear and consistent with the existing literature on self talk. The the experimental group improved significantly more from baseline to the final assessment and outperformed the control group at the endpoint of the study. This replicates what prior research has established many times over and confirms that the self talk intervention was effective, but the more novel contribution of this paper lies in the HRV analysis. At the final assessment, but not at baseline, the experimental group showed a different pattern of RMSS deactivation across the phases of the task compared to the control group. [00:09:37] Specifically, the self taught group showed progressively higher parasympathetic activation in the later stages of the pudding task, a pattern that was not present before the training intervention and was not observed in the control group at either time point. The interpretation the authors offer is compelling. Self talk training accelerated the development of more automatic, less effortful attentional processing, and as the task continued within a session, the self talk group's nervous system settled into increasingly efficient, parasympathetically dominated regulation. [00:10:12] The control group, still relying more on effortful conscious control, did not show this convergence toward calm, efficient processing. There is an important caveat that I want to address directly because the authors themselves are careful to emphasize it, and I think intellectual honesty demands that we honor it. [00:10:32] The HRV findings in this study did not reach multivariate statistical significance in the formal omnibus tests. What this means practically is that while the described patterns were present and interpretable in the data, the study did not have sufficient statistical power to conclude definitively that self talk caused the observed HRV differences. [00:10:53] The authors used the word indications deliberately. [00:10:56] These are suggestive findings that provide a plausible mechanistic account and a clear direction for future research. Not confirmed effects with 40 participants and the inherent variability in HRV during an active motor task, the study was not powered to detect small to moderate autonomic effects with high confidence. This limitation does not undermine the study's value. It is exactly the kind of honest, calibrated reporting that good science requires. But it does mean that the HRV findings should be held lightly pending larger and more rigorous investigations. If future studies with larger samples and more sophisticated psychophysiological designs replicate these patterns, the theoretical framework this paper offers will become a significant contribution to our understanding of how mental skills training operates at the level of the autonomic nervous system. The broader context for this study is worth appreciating. [00:11:51] Sports psychology has long argued that the mental skills that elite athletes develop concentration, confidence, emotional regulation, and yes, self talk are not separate from physiology but deeply intertwined with it. The autonomic nervous system is the biological substrate through which mental states manifest as physical states and vice versa. Anxiety produces cardiac acceleration and suppressed vagal tone focused calm produces slower heart rate and higher hrv. [00:12:22] The fact that a psychological intervention like self talk training might produce detectable autonomic changes is not surprising from a theoretical perspective. What has been missing is the empirical demonstration done rigorously that these changes are real and related to performance. [00:12:40] This study takes a meaningful step toward that demonstration. [00:12:44] The researchers are also thoughtful about what they call the temporal dynamics of the HRV response within the task. They did not just measure average HRV across the entire session, but looked at how it changed as the task progressed. This temporal sensitivity is important because the most interesting theoretical prediction is not that self talk produces uniformly higher RMSSD throughout performance, but that it accelerates the shift toward more efficient automatic processing as the task continues. And that is what the data tentatively suggests. [00:13:21] Future research that tracks HRV in finer temporal resolution during performance and that relates those fine grained autonomic dynamics to moment to moment attentional state and performance quality will be building directly on the foundation this study lays. [00:13:38] Our second study today moves from the pudding green to what might as well be a nap lab and asks a fundamentally different question about HRV and performance. [00:13:47] Can we detect the physiological correlates of mental fatigue using a completely unobtrusive sensor system and specifically one that requires no attachment to the body whatsoever? [00:13:59] The full title is Exploratory Temporal Dynamics of Heart Rate Variability during a Prolonged two back Task under Nap Deprivation Conditions, and the authors are Diming Lu Yaoyu, Tianjiaoming, Jing Jiang, Liu Fengzhou, Hengchong, Gong Yifeng, and Ying He. Working from the Naval Medical center of Naval Medical University in Shanghai, the team's background in military medicine and human factors research gives the study a clear practical motivation. [00:14:31] Understanding cognitive fatigue in operational environments where wearing sensors may be impractical, uncomfortable, or logistically impossible. The technology at the heart of this study is the ballistocardiogram, abbreviated bcg, and if you have not encountered it before, it is worth explaining in some detail because it represents a genuinely different approach to cardiac monitoring. Every time the heart contracts and pumps blood, the force of that ejection transmits mechanical energy to the body. [00:14:59] The recoil from each heartbeat is tiny but measurable. It produces small movements of the entire body that propagate from the heart outward through the tissues and joints. If you place a sufficiently sensitive force or displacement sensor in contact with the body, you can capture these micro movements and use them to estimate the timing of each heartbeat. From those timing estimates, which serve as surrogates for the R to R intervals you would get from a standard electrocardiogram, you can compute heart rate variability metrics. The researchers in this study used a fiber optic sensing cushion that participants simply sat on during the task with no attachment, no gel, no electrodes, and no conscious awareness of being monitored. The practical appeal of this technology is enormous. A sensor embedded in a seat cushion or a car seat or an office chair or an airplane seat could potentially provide continuous autonomic monitoring without any burden to the person being monitored. The theoretical applications are vast, detecting fatigue in pilots, monitoring stress in office workers, tracking cognitive load in military operators, or identifying drowsiness in truck drivers. But the BCG signal is considerably noisier than a direct ECG recording, particularly when the subject is moving, and the accuracy of the derived heart rate and HRV metrics depends heavily on signal quality and motion control. The researchers were working at the frontier of what this technology can currently do. 26 healthy graduate students participated in the study. They completed a 60 minute version of the two back task, a classic working memory paradigm in which participants see a sequential stream of stimuli and must indicate whether each one matches what appeared two positions earlier in the sequence. The task demands continuous updating of working memory, sustained attentional focus, and precise response inhibition. It is cognitively taxing under any circumstances and completing 60 minutes of it during the period of the day when each participant would normally nap added the additional challenge of fighting circadian rhythm driven sleepiness. The task was timed to coincide with each participant's habitual nap window, a period associated with a natural trough in alertness for many individuals. 22 of the 26 recordings were usable for HRV analysis after 4 were excluded due to insufficient compliance with the required seated posture protocol, reflecting one of the practical limitations of BCG technology. [00:17:33] Even moderate postural shifts introduce motion artifacts that degrade signal quality. The cognitive and fatigue effects of the task were substantial. [00:17:43] Scores on the mental fatigue subscale of the Fatigue Self Assessment Scale rose from approximately 19 points before the task to approximately 47 points afterward, a change of approximately 28 points with a Cohen's D of 1.32, which is by any standard a very large effect. Reaction time on a simple cognitive assessment also slowed meaningfully by approximately 36 milliseconds with an effect size of 0.98. [00:18:11] So the behavioral and self report evidence for significant mental fatigue following 60 minutes of demanding cognitive work during the nap deprivation period was unambiguous and strong. The question was whether the 19 HRV features computed from the BCG derived interbeat interval data analyzed in rolling 5 minute windows with 1 minute step increments across 4 descriptive phases of the task would show systematic changes reflecting this fatigue. [00:18:38] The short answer is that they did not, at least not at the group level. After applying the appropriate statistical corrections. When the researchers used a false discovery rate correction to account for the fact that they were testing 19 different HRV features simultaneously, none retained statistical significance patterns that appeared promising before correction vanished once the multiple testing burden was properly controlled. [00:19:06] But this null result is not a simple failure. It is, I would argue, one of the most informative findings in the entire study. [00:19:15] When the researchers took an exploratory look at individual level HRV trajectories across the task, the picture that emerged was one of striking heterogeneity. [00:19:25] Different participants showed radically different autonomic patterns over the 60 minute period. [00:19:31] Some showed HRV features that declined as fatigue accumulated. Others showed increases. Others showed non monotonic patterns with peaks and valleys that did not align with any simple linear fatigue narrative. When such varied, individual responses are averaged together at the group level, they cancel each other out, leaving a flat, apparently uninformative group mean. [00:19:55] The lesson here is not that HRV cannot reflect cognitive fatigue it may well do so at the individual level. [00:20:03] The lesson is that cognitive fatigue does not produce a uniform universal HRV signature that is reliably detectable across all individuals and that group level analyses may be fundamentally the wrong tool for the job of detecting individual level physiological states in real time. This has profound practical implications for anyone hoping to build an HRV based cognitive physical fatigue monitoring system, whether for industrial safety, military applications, or clinical practice. A monitoring algorithm trained on group average patterns and applied to individuals may simply not generalize because the individual level signal varies so dramatically from person to person. [00:20:50] Training and calibrating monitoring systems on each individual's own baseline and response patterns may be essential for making HRV based cognitive fatigue detection work in practice. [00:21:03] This is a hypothesis, not a conclusion, but it is one that the heterogeneous trajectory data in this study strongly motivates. [00:21:12] It is also worth saying something about the false discovery rate correction and why it matters, because I know some listeners may not be familiar with the statistical reasoning. When you test 19 different hypotheses at once, does this HRV feature change? Does that one change? And so on for each of the 19 metrics, the probability of finding at least one spuriously significant result just by chance accumulates rapidly. If you use a standard significance threshold of 5% for each test, the expected number of false positives across 19 tests is nearly one. The false discovery rate correction adjusts the required significance threshold for each individual test to control the overall rate of false discoveries across all tests together. [00:22:02] It is the responsible thing to do when analyzing multiple endpoints, and the researchers applied it here, which is commendable. Many studies in this area do not. The consequence is that patterns which appeared interesting without correction and disappeared with it, and rather than treating this as a defeat, the researchers correctly frame it as information. [00:22:24] The HRV signal during this cognitive task analyzed at the group level does not contain a reliable group wide fatigue marker of sufficient magnitude to survive statistical scrutiny. That is a genuine and useful finding. The implications for BCG technology specifically are nuanced. [00:22:44] The researchers found that the cushion sensor produced usable HRV data in 22 of 26 participants, which is a reasonable success rate, but also means roughly 15% of recordings were unusable due to motion artifacts. In a real world deployment context, an office building, a vehicle, an airplane, the percentage of unusable recordings might be higher because real world environments impose more motion and postural variability than a controlled cognitive task protocol. Before BCG based HRV monitoring can be deployed at scale, the signal processing algorithms that extract clean interbeat intervals from the noisy mechanical signal will need to be substantially more robust. [00:23:29] Progress is being made on this front. Machine learning based artifact rejection and signal reconstruction approaches are becoming more sophisticated, but the gap between laboratory feasibility and reliable real world performance remains real and significant. [00:23:46] Before we continue, a quick word about Optimal HRV if you're serious about understanding your heart rate variability and want to track it with tools that are both scientifically rigorous and genuinely accessible, I want to invite you to explore the Optimal HR V app and the suite of sensors we support. [00:24:05] We built optimal HRV because we believe that the science of autonomic nervous system monitoring should not be locked away in research laboratories. It should be in the hands of athletes, clinicians, coaches, and curious individuals who want to make better decisions about their health and performance. [00:24:24] The app makes it straightforward to measure your HRV consistently, track how it changes over time, understand what those changes mean in context and and use that information to guide your training, recovery and stress management. [00:24:38] Visit optimalhrv.com to learn more and check the show notes for a direct link to get started. [00:24:45] Now back to the science. [00:24:47] Our third study today takes us from cognitive laboratories and sensor cushions to a driving simulator and asks a question with immediate and serious implications for public safety Can HRV tell us when a driver is dangerously drowsy? [00:25:03] The study is titled Tracking Vigilance While Pilot Study of Heart Rate Variability Classification under a Controlled Sleep Deprivation Protocol, and the authors are James Albert Duverger, Eli Moser, Philippe Boudreau, Marie Claude Ouimet, DNB Boivin, and Alireza Saidi. The team spans three the Institut de Recherche Robert Sauve Enceinte Travail in Montreal, the center for Study and Treatment of Circadian rhythms at McGill University's Douglas Mental Health University Institute, and the Faculty of Medicine and Health Sciences at the University of Sherbrooke. It is a genuinely interdisciplinary collaboration combining occupational health research, circadian biology, traffic safety expertise, and biomedical engineering. [00:25:56] Exactly the kind of team you need to attack a problem this complex. [00:26:00] The problem they are tackling has enormous stakes. [00:26:04] Drowsy driving contributes to a disproportionate share of the most severe traffic crashes, those involving fatalities and serious injuries. [00:26:13] Research from the AAA foundation for Traffic Safety in the United States estimates that drowsiness may be a contributing factor in between 16 and 21% of fatal crashes, and similar figures appear in data from Canada, Europe, and Australasia. What makes drowsy driving particularly difficult to address is the self assessment problem. Drivers who are severely sleep deprived consistently underestimate how impaired they are. The feelings of sleepiness come and go in waves, and during the brief moments of relative alertness that punctuate severe drowsiness, a driver may genuinely believe they are fine to continue. [00:26:53] Behavioral cues, lane deviation, steering corrections, eyelid droop only become reliably detectable after impairment is already severe enough to pose an acute crash risk. Physiological monitoring offers a more direct route to impairment detection because it measures the internal state of the nervous system rather than waiting for the behavioral consequences of that state to manifest. [00:27:17] The autonomic nervous system regulates alertness and arousal, and sleep deprivation disrupts those regulatory systems in ways that change the statistical properties of the cardiac autonomic signal. HRV features that reflect the balance between sympathetic and parasympathetic tone, the complexity of cardiac regulation, and the coherence of oscillatory patterns in the heart rate signal are all candidates for sensitivity to the drowsiness induced changes in all autonomic state. The study recruited 12 healthy young adults and put them through a 30 hour constant routine sleep deprivation protocol. The constant routine is a rigorous laboratory paradigm designed specifically to isolate the effects of sleep deprivation from the confounding influences of circadian phase, physical activity, posture, ambient temperature, and food intake. Participants are kept in identical control conditions throughout the protocol and same lighting, same posture, same meal timing, same physical activity level, and driving. Simulations were administered every two hours across the entire 30 hour period. [00:28:25] At each session, ECG data, self reported sleepiness ratings using a validated scale, psychomotor vigilance test performance, and driving performance metrics were collected. The primary analysis compared the very first session the rested state to the very last session the tired state, after approximately 30 hours of continuous wakefulness. This is a clean, well powered experimental contrast designed to maximize the detectable hrv. Differences between states and the differences between states were exactly as expected. [00:29:00] Subjective sleepiness ratings were dramatically elevated in the tired state. Psychomotor vigilance performance deteriorated substantially. [00:29:09] Driving variability sees a metric reflecting the degree of compensatory steering corrections needed to maintain lane position increased significantly, indicating impaired vehicle control. [00:29:21] The two states were unambiguously different at the behavioral and subjective levels, which is the necessary foundation before asking whether HRV can distinguish them. [00:29:31] Multiple HRV features showed statistically significant differences between the rested and tired states, with effect sizes ranging from moderate to large. This is consistent with the existing literature on sleep deprivation and autonomic function, which shows that prolonged wakefulness reorganizes the balance of autonomic regulation, reduces parasympathetic tone, increases sympathetic influence, and alters the complexity and variability of the cardiac signal in ways that reflect degraded homeostatic capacity. [00:30:03] The HRV signal was sensitive to these changes and the magnitude of the differences was practically meaningful, not just statistically detectable. The machine learning classification analysis is the methodological highlight of this paper. The researchers trained a set of classifiers to predict whether a recording came from the rested or tired state using HRV features as the input variables. The key design choice was using leave one subject out cross validation. [00:30:31] Each classifier was trained on data from 11 participants and tested on the held out 12th with the process repeated for each participant. This ensures that the reported accuracy reflects how well the classifier generalizes to individuals it has never seen before. A much more demanding test than simply fitting a model to all the data and checking how well it fits. All classifiers evaluated in the study achieved average accuracy greater than 85% under this stringent cross validation scheme. That is a compelling result. The interpretation is that the HRV Signature of 30 Hour Sleep Deprivation Generalizes across individuals to a degree that supports accurate participant independent classification. [00:31:17] Despite the individual variability in baseline HRV that complicates so many other analyses. The effect of this level of sleep deprivation on the HRV signal was considered enough across participants to remain detectable even when the classifier encountered a new individual it had never seen. [00:31:35] This is an important positive result for the field of HRV based physiological monitoring in safety critical context. The authors are honest about the limitations and I want to make sure we convey them accurately. 12 participants is a small sample. The entire cohort consisted of healthy young adults and and total 30 hour sleep deprivation may be more extreme than the cumulative partial sleep loss that characterizes real world fatigue, driving in shift workers, commercial drivers, and the general population. The laboratory environment removed many confounds that would complicate real world HRV measurement. Road vibration, thermal stress, emotional demands of real traffic police, postural changes, and the two state classification problem studied here, rested versus severely sleep deprived may be considerably easier than the continuous real time detection problem that a deployed drowsiness monitoring system would face. [00:32:36] None of these limitations diminish the contribution they clarify the path forward. [00:32:42] The study provides a methodologically rigorous foundation and a clear demonstration of principle, and the more than 85% classification accuracy gives the field a meaningful performance benchmark to build toward and eventually exceed. The specific HRV features that proved most sensitive to the rested versus tired contrast are worth discussing, even though the study's primary contribution is the classification result rather than the individual feature analysis in the drowsy driving literature. Broadly, features related to parasympathetic modulation, which declines with sleep deprivation as sympathetic tone rises, have been among the most consistent discriminators. Time domain metrics like RMSSD and the percentage of successive differences exceeding 50 milliseconds tend to decline with increasing sleep loss. In the frequency domain, the high frequency component of hrv, which reflects respiratory linked vagal modulation, also tends to decline, while the ratio of low frequency to high frequency power may increase, reflecting the relative shift toward sympathetic dominance. [00:33:52] Nonlinear metrics that capture the complexity and multiscale structure of the heart rate signal, including sample entropy and detrended fluctuation analysis measures, have also shown sensitivity to sleep deprivation in the existing literature. This study does not report features feature by feature results in the level of detail that would allow us to say definitively which features were most discriminating in this specific dataset. But the overall pattern of results is consistent with what prior research has found for the long term project of deploying HRV based drowsiness detection in commercial vehicles. There are additional engineering challenges beyond those the study directly addresses. Obtaining a clean ECG signal in a vehicle requires either chest electrodes, which most drivers would not accept, or steering wheel sensors capable of detecting the cardiac signal through the driver's hands. Capacitive sensing arrays in seats, similar in concept to the BCG cushion in the NAP study are another possibility. [00:34:53] The noise environment of a moving vehicle vibration, road irregularities, the electrical interference of the vehicle's own systems is substantially more challenging than the control laboratory. [00:35:06] The HRV features that discriminate rested from tired states in the laboratory may not be extractable with sufficient reliability from a vehicle mounted sensor, or they may require signal processing approaches that introduce unacceptable latency for a real time warning system. [00:35:24] None of these challenges are insurmountable, but they are the difference between a compelling laboratory demonstration and a deployable product, and they deserve serious attention in the next generation of research. [00:35:36] Our fourth and final study today addresses a question that I think is one of the deepest and most underappreciated in applied HRV science. When you apply the same training to a group of athletes, do they all respond the same way? The study is titled Identification and Longitudinal Monitoring of Physiological Response Profiles in Wrestlers using Multivariate Heart Rate Variability Analysis, authored by Galya Georgieva Tsaneva of the Institute of Robotics at the Bulgarian Academy of Sciences in Sofia. [00:36:11] The Institute of Robotics is not a place you would immediately associate with sports science research, but the intersection of biomedical signal processing, pattern recognition and human physiology is exactly where computational approaches like the ones used here originate, and the methodological sophistication of this single author paper reflects that background the study recruited 65 competitive wrestlers and followed them across five measurement points over a four month training period. [00:36:43] At each measurement point, Holter electrocardiogram recordings were obtained before and immediately after a standardized training session, providing a measure of both baseline autonomic state and and the acute autonomic response to training stress at the individual level. The HRV analysis was comprehensive time domain metrics including root mean square of successive differences and standard deviation of normal to normal intervals or SDNN frequency domain metrics including the low frequency to high frequency ratio point array plot indices including SD1, the measure of short term beat to beat variability, and SD2, the measure of longer term variability, reflecting slower oscillatory patterns and nonlinear complexity metrics including sample entropy, which quantifies the unpredictability and irregularity of the cardiac rhythm over time. Each of these metrics captures a different dimension of autonomic function and together they describe the state of the cardiac autonomic control system in a richer and more multidimensional way than any single metric alone. For each athlete at each measurement point, the researchers computed the percent change from before to immediately after training for every HRV metric. These percent change vectors represent each individual's acute autonomic response profile, not what their resting HRV looks like in isolation, but how their autonomic system behaves when it is confronted with a standardized training load. The analysis proceeded through two stages. [00:38:21] First, principal component analysis was applied to the response vectors to reduce the dimensionality of the Data. [00:38:28] The first two principal components together explained 80.53% of the total variance, a remarkable result indicating that despite starting with many HRV variables, the underlying structure of how athletes respond to training can be captured almost entirely in just two dimensions. Then hierarchical clustering was applied in this two dimensional space to identify natural groupings of athletes with similar response profiles. Two distinct clusters emerged with striking consistency. Profile a containing 55 athletes. The substantial majority of the sample showed what the literature describes as a classical training response. During and immediately after the training session, heart rate increased and HRV metrics declined reflecting a shift towards sympathetically dominated autonomic regulation as the body mobilizes its resources for physical performance. [00:39:24] Rmssd, sdnn, sample entropy in the poinche indices all decreased and the low frequency to high frequency ratio shifted in ways consistent with increased sympathetic and decreased parasympathetic influence. This is the expected canonical acute response to intensive exercise and seeing it in the majority of athletes confirms the validity of the measurement approach. [00:39:49] Profile B containing the remaining 10 athletes showed essentially the opposite pattern. Their HRV indices were maintained or in some cases elevated following training, suggesting that their autonomic systems did not experience the same degree of sympathetically driven suppression in response to the identical training session. The two profiles were most dramatically differentiated by their responses in RMSSD, SD and the low frequency to high frequency ratio with sample entropy and the Peinchere measures also contributing meaningfully to the separation. [00:40:25] The fact that two such distinct profiles emerged from the same sample of athletes doing the same training is itself a striking finding, but the longitudinal data are what transform it from a curious cross sectional observation into a potentially important scientific finding. [00:40:44] Across all five measurement points over the four month training period, the profiles were reproducible. [00:40:50] Profile A wrestlers consistently showed the sympathetically dominated response at each assessment. [00:40:57] Profile B wrestlers consistently showed the preserved or elevated HRV response. The profiles were not random variation from session to session. They were stable individual level patterns in how each wrestler's autonomic system responded to training stress and they persisted over a period of months. [00:41:17] This stability is what makes the finding scientifically meaningful. It suggests the profiles reflect something real and reproducible about individual differences in autonomic reactivity to exercise, not just measurement noise or day to day physiological fluctuation. [00:41:34] What might explain Profile B? The authors appropriately leave several possibilities open. These athletes might be exceptionally well adapted to the their training load, so highly trained that the session no longer constitutes the same relative physiological challenge it poses to Profile A athletes. They might be entering training in a different physiological state, better rested, lower in accumulated fatigue with higher baseline parasympathetic tone such that the same absolute training load produces a different relative autonomic impact. Or there might be a constitutional difference in autonomic reactivity that is independent of training adaptation and represents a stable individual trait. Distinguishing between these explanations would require additional data, measures of training load relative to each athlete's capacity, assessments of recovery status, and comparisons of long term performance outcomes between profiles. The current study cannot adjudicate between these possibilities and the authors are careful not to upgrade Overclaim the practical significance for coaches and sports scientists is substantial. If a substantial minority of athletes on a given training program are showing a fundamentally different acute autonomic response profile than the majority, one that looks like the opposite of what is expected, then monitoring based on group averages will correctly characterize neither group. A coach reviewing aggregate HRV trends for a squad might see a stable or slightly declining average that masks the fact that 10% of athletes are operating in a completely different physiological regime, one that may require different training, recovery or monitoring strategies. The methodology demonstrated here, collecting before and after HRV at each training session, computing percent change response vectors, and applying multivariate dimensionality reduction in climate clustering, gives coaches and sports scientists a rigorous, data driven way to surface this kind of heterogeneity and act on is not a trivial undertaking. Holter monitoring is not as simple as a wrist sensor, but the analytical approach is clear and replicable, and the finding that two meaningful profiles exist challenges us to take individual response heterogeneity seriously as a practical matter in athlete monitoring. [00:43:57] I also want to highlight the study's use of sample entropy and Poincheur a NDC's because these metrics are less familiar to many HRV practitioners than RMSSD and are worth understanding. Sample entropy measures the irregularity or unpredictability of the beat to beat interval sequence over time. [00:44:16] A higher sample entropy generally indicates more complex, less predictable cardiac regulation, which tends to be associated with healthier, more adaptive autonomous function. After intensive training, sample entropy typically declines as the autonomic system is stressed and its regulatory complexity is temporarily reduced. The Poincare plot is a geometric representation of HRV in which each beat interval is plotted against the interval that follows it. SD1, the width of the ellipse perpendicular to the line of identity reflects short term vagal modulation of the heart rate closely related to RMSSD. [00:44:53] SD2, the length of the ellipse along the line of identity reflects longer term variability driven by slower autonomic oscillations and overall variability in heart rate. Together, SD1 and SD2 provide a decomposition of total HRV into its short term and longer term components, which carry somewhat different physiological interpretations. The Ressler study used all of these metrics together and the result was a richer pair picture of individual response patterns than any single metric could have provided in isolation. The Ressler study also raises an important question about the appropriate time scale for HRV monitoring in competitive athletes. Much of the HRV monitoring literature focuses on morning resting HRV as a readiness indicator measured once per day under standardized conditions before training. This approach has real value in a solid evidence base, but it captures only one dimension of the HRV story what happens to HRV during and immediately after training? The acute response profile is a different and complementary kind of information. [00:46:00] It tells you not just where an athlete starts their day, but how their autonomic system responds when challenged. An athlete who has high resting HRV but shows a blunted acute response to training might be in a different physiological state than one with the same resting hrv, but a robust acute response. The before and after design used in the Ressler study is not something that can be easily replicated with consumer wearables in a real training environment, but it points toward a richer conception of what comprehensive HRV monitoring could look like for competitive athletes, not just daily resting values, but a more complete picture of autonomic dynamics across the training day. The choice to use competitive wrestlers also matters for the generalizability of the findings. Wrestling is a sport that combines aerobic fitness, anaerobic power, maximal strength, technical skill, and considerable psychological stress. It is in many ways a demanding whole body athletic context that is not easily reduced to a simple fitness variable. The diversity of the training demands means that the HRV response profiles identified here may reflect a genuinely complex mix of physiological, psychological and technical factors rather than a single underlying dimension. Understanding which of these factors drive the Profile A versus Profile B distinction would require a richer data collection than a a single study can provide. But it is exactly the kind of question that longitudinal athlete monitoring programs, increasingly common in elite sport, are positioned to answer. One additional finding worth emphasizing is the explanatory power of the first two principal components. The fact that 80.53% of the total variance in the multivariate HRV response data was captured in just two dimensions tells us something important about the structure of the data. It suggests that despite the apparent complexity of tracking 19 HRV metrics simultaneously, the underlying biology is producing a relatively low dimensional response. [00:48:09] Athletes are not varying independently on each of the 19 metrics in arbitrarily different ways. They are varying systematically along a small number of underlying dimensions, which is why PCA can compress so much of the variance and into just two components. This structural regularity is what makes the clustering approach work. If the response space were truly high dimensional, with participants scattered throughout a vast 19 dimensional cloud, no simple clustering approach would find meaningful groupings. The low dimensionality of the response space revealed by PCA is what allows the two profiles to emerge cleanly and be distinguished reproducibly across measurement points. [00:48:53] Stepping back and looking across all four studies, the picture that emerges is one of a field grappling productively with the complexity of individual differences. [00:49:03] In the Self Talk study, the HRV response to attentional training showed different patterns between the experimental and control groups, but was accompanied by substantial within group variation that limited statistical confidence. [00:49:16] In the cognitive fatigue study, the most striking finding was not a clear group level HRV marker of mental fatigue, but the opposite highly heterogeneous individual trajectories that resisted any clean group level summary and suggested that personalized monitoring approaches may be essential. [00:49:34] In the drowsy driving study, impressive group level classification accuracy was achieved, but the participant independent validation validation design itself acknowledges that some individuals may be harder to classify than others, and the study's limitation to healthy young adults leaves open the question of how well the approach generalizes across the population of interest. [00:49:57] And in the Ressler study, the entire scientific contribution is built on demonstrating that meaningful individual differences in HRV response profiles exist, exist, are reproducible over time, and would be missed by aggregate analyses. There is also a shared theme across these four papers of HRV as a window into internal states that are otherwise invisible. The attentional shifts that Self Talk produces cannot be directly observed from the outside. [00:50:26] HRV offers a potential physiological trace. The experience of cognitive fatigue is subjective and self reported. [00:50:34] HRV from a sitting cushion offers a potential objective correlate. The vigilant state of a drowsy driver is dangerous precisely because it is not apparent from behavioral observation until it is too late. HRV offers a potential early detection signal. The adaptive heterogeneity of wrestler responses to training is buried in individual physiological data that aggregates statistics arrays. [00:51:00] Multivariate HRV profiling surfaces it across all four contexts. HRV is functioning as an interpreter, translating the biology of the autonomic nervous system into information that can inform decisions in sport safety and clinical practice. That interpretive role is why HRV research matters, and it is why this show exists. What remains genuinely uncertain is is as important to name as what has been established. We do not yet know whether the attentional HRV changes observed in the GOLF study will replicate with larger samples and more rigorous psychophysiological designs. We do not know which HRV features, if any, will prove to be reliable individual level markers of cognitive fatigue in real operational environments. The NAP study's null group results should make us appropriately humble about claims in this space. [00:51:55] We do not know how well the drowsy driving classifier will perform in older populations, people with sleep disorders or people experiencing the chronic partial sleep loss that characterizes real world shift work rather than acute total deprivation in the laboratory. And we do not know what causes Profile B in wrestlers, whether it is beneficial or concerning, or whether it requires different training management. [00:52:22] Each of these open questions is not a failure of the field, it is a map of where the next generation of research needs to go. A fourth theme I want to highlight is the question of measurement fidelity and what it means for clinical translation. Every study we reviewed today was working at or near the limits of current measurement technology. [00:52:42] The BCG cushion in the NAP study was a frontier technology which with real signal quality limitations. [00:52:49] The ECG monitoring in the Drowsy Driving study was laboratory grade but would face substantial challenges in vehicle deployment. The holder recordings in the Ressler study are far more detailed than what consumer wearables provide. The Self Talk study used a standard sports psychology HRV protocol that captured useful information but could not resolve the fine grained temporal dynamics that would most directly test the attentional mechanism hypothesis. In each case the researchers were doing the best that current technology and methodology allow, and in each case there are clear directions for improvement that better technology and larger samples would enable. This is the normal condition of a healthy scientific field and it should be understood as productive tension rather than systematic failure. The translation from research findings to practical tools is also worth reflecting on. HRV has already made the transition from research laboratory to consumer device in a meaningful way. Millions of people now track their HRV with wrist based optical sensors, smart rings and chest straps and the data they collect is real, physiologically meaningful, and in many cases actionable. [00:54:05] But the gap between what the research demonstrates in controlled conditions and what consumer devices reliably measure in everyday life remains significant. [00:54:15] The HRV metrics that showed the clearest effects in today's studies were computed from high quality ECG or Holter recordings. Consumer optical sensors have lower signal to noise ratios and may not capture the more nuanced noise. Nonlinear and frequency domain features as reliably as device hardware and signal processing software continue to improve and they are improving rapidly. This gap will narrow and the clinical and performance monitoring applications suggested by today's research will become increasingly practical. [00:54:52] I also want to say something about the value of exploratory and null results because two of today's four studies produce findings that were null or only partially significant, and in the HRV field there is sometimes a tendency to discount those kinds of results and the NAP study found no significant group level HRV markers of cognitive fatigue. After statistical correction, the Self Talk study found HRV effects that were suggestive but did not reach formal significance. Both papers were published in peer reviewed journals. Despite these null or limited results, results which reflects good editorial judgment, null results that are properly powered, properly analyzed and honestly reported are genuinely valuable. They tell us where the signal is not, which is often as important as knowing where it is. A field that only publishes positive results quickly becomes distorted in its understanding of what interventions and measurements actually do, and the corrective process of revisiting and failing to replicate results in larger, better powered states studies is painful and slow. The researchers in these two studies did the right thing by reporting their findings honestly, even when those findings were more complicated than a simple positive result would have been. [00:56:03] There is one more thread I want to pull on before we close because I think it is underappreciated in how we talk about HRV monitoring in practice. All four studies today involve HRV as a dependent variable, something that changes in response to interventions, states and conditions. [00:56:21] This is the framing that dominates the research literature. We manipulate something and we measure HRV to see what happened. But for practitioners, coaches, clinicians, athletes, occupational health professionals, HRV is typically used as an independent variable, a predictor of something else. Readiness to train risk of illness, cognitive capacity, drowsiness level the translation from the research framing to the practitioner framing is not automatic. A study showing that self talk training changes HRV does not by itself tell you whether monitoring HRV in real time during performance will help you understand what an athlete is doing intentionally. [00:57:04] A study showing that HRV features differ between rested and severely sleep deprived drivers does not by itself tell you whether a wristband HRV reading taken at the start of a shift will predict crash risk reliably enough to justify action. These are different questions, and they require different research designs than the studies reviewed here. The work that bridges laboratory demonstrations of HRV sensitivity with real world monitoring utility remains largely ahead of us, and it will require the collaboration of physiologists and engineers, statisticians, ethicists, and the people whose health and safety are ultimately at stake. [00:57:44] What I take from all of this and what I hope you take from today's episode is a picture of HRV science that is genuinely exciting, scientifically honest, and appropriately humble. The field is producing results that matter using increasingly creative methodological approaches and being honest about what those results do and do not assess. Establish the Self Talk study opens a new window into the autonomic correlates of psychological skills training. The Cognitive Fatigue study identifies the limits of group level HRV analysis and makes a case for personalized monitoring approaches. The Drowsy Driving study establishes a performance benchmark for HRV based physiological state classification that the field can build towards. [00:58:29] And the Ressler study provides a methodological roadmap for characterizing individual response heterogeneity in ways that could transform how we think about athlete monitoring. Each of these contributions is real. Each of them also points honestly toward what needs to happen next. That combination genuine progress paired with genuine humility is what a healthy scientific field looks like. [00:58:56] One practical note before I sign off. If you are a practitioner using HRV monitoring, whether with athletes, patients, or your own data, today's episode reinforces something I come back to regularly on this show. Individual context matters more than group norms. Your own HRV baseline, your own response to training, your own patterns of fatigue and recovery are more informative for your decisions than population average. [00:59:22] The Ressler study showed that even within a relatively homogeneous group of competitive athletes doing the same training, individual response profiles diverged dramatically. The Cognitive Fatigue study showed that individual trajectories of HRV during sustained mental work can point in opposite directions for different people. The message across all four papers is the same. The population average is a starting point, not a destination. [00:59:48] Your own longitudinal data, collected consistently over time and interpreted in the context of your life and goals is where the real insight lives. [00:59:58] I want to return briefly to the methodological theme of this episode because I think it is worth making explicit. Three of the four studies today used relatively small sample sizes 40 participants in the golf study, 22 usable recordings in the cognitive task study, and 12 participants in the Drowsy Driving study. Only the Ressler study had a sample size 65 athletes large enough to support the kind of multivariate analysis the researchers wanted to do, and that study was explicitly designed around the need for adequate sample size to support principal component analysis and hierarchical clustering. [01:00:37] In the other three studies, the small samples impose real limits on what can be concluded and and what effect sizes can be reliably detected. This is not a unique problem in HRV research. It reflects the broader challenge that collecting high quality physiological data from human participants is expensive, time consuming, and ethically constrained. [01:00:59] Each participant in the Drowsy driving study spent 30 hours in a controlled laboratory environment and underwent repeated driving simulations. [01:01:08] Each participant in the golf study completed four sessions of putting practice with physiological monitoring. The cost and logistical burden of this kind of research inherently limits sample sizes. Understanding those limits and not interpreting findings beyond what the statistical power supports is essential for reading the literature accurately. [01:01:29] The honest framing for most of today's findings is these are promising, plausible, theoretically coherent results that establish the feasibility of the proposed approach and warrant investigation in larger, better powered studies. [01:01:43] That framing is accurate, scientifically appropriate, and genuinely exciting because it means the field is in an active and productive phase of hypothesis generation and preliminary validation. [01:01:55] The geographic and demographic diversity of today's studies is also worth noting because it suggests the HRV field is genuinely global. The Self Talk study came from a Greek research team, the Cognitive Fatigue study from China, the Drowsy Driving study from Canada, the Ressler study from Bulgaria. Each of these contexts brings different training cultures, different populations, and different research traditions to bear on the common question of what HRV can tell us about human performance. [01:02:27] Cross cultural replication Seeing whether findings in one population generalize to others is one of the ways science builds confidence in its conclusions, and the international spread of HRV research makes that kind of replication increasingly possible. [01:02:43] Thank you so much for being here for today's episode of this week in hrv. It genuinely means a lot to me that you take the time to engage with this science, whether you're listening on your morning run and your car car at your desk, or in whatever context your life brings you. The HRV research community is producing more and better work than ever before, and being able to share it with an audience that is curious and engaged is one of the most rewarding things I do. If you found any of these studies useful, whether it changed how you think about self talk and the nervous system, gave you a more nuanced understanding of what HRV can and cannot tell us about. Cognitive fatigue opened your eyes to the possibility of physiological drowsiness detection, or challenged you to think more seriously about individual differences in how athletes respond to training. I would love it if you would share this episode with someone in your life who would appreciate it. A colleague, a coach, a clinician, a training partner, whoever in your circle is curious about this stuff. That is how science reaches the people who can use it. [01:03:49] All four studies discussed today are linked in the show notes along with the full author information, key findings, and significance summaries for each one if you have questions, reactions or study suggestions, and especially if you have data or clinical observations that speak to any of the themes we discussed today, please reach out through the Optimal HRV website or find us on social media. [01:04:13] I am Matt Bennett and I will see you next week for more of this week in hrv.