Heart rate variability coefficient of variation during sleep as a digital biomarker that reflects behavior and varies by age and sex, 2026, Grosicki

[Mij]

NEW & NOTEWORTHY We identify HRV-CV measured during sleep as a scalable measure of day-to-day HRV fluctuation. Analyzing nearly 2 million HRV readings from >21,000 individuals, we document that five nights are sufficient for reliable 7-day estimates. Higher HRV-CV is associated with greater alcohol consumption, lower activity, shorter and less consistent sleep, older age, and higher BMI.
These findings provide the first large-scale characterization of HRV-CV and support its potential as a biomarker for behavioral monitoring and risk stratification.

Abstract
The heart rate variability coefficient of variation (HRV-CV) is an index of day-to-day cardiac autonomic fluctuation that may serve as a scalable digital biomarker for behavioral monitoring and health risk stratification. We investigated how many nights of sleep-derived HRV were needed to reliably estimate 7-day HRV-CV, and examined associations with behavioral and demographic characteristics linked to health. We analyzed ∼2 million nocturnal HRV readings from >21,000 wearable device users, stratified by age and sex. Seven-day HRV-CV was calculated, and a simulation determined the minimum nights required for reliable estimates.

Associations with alcohol, physical activity, sleep, and variability of these behaviors were evaluated. Additional models examined associations between HRV-CV with age, biological sex, and body mass index (BMI). At least five of seven nights were required to achieve acceptable agreement with full-week HRV-CV values (intraclass correlation coefficient ≥ 0.80). Higher HRV-CV was associated with greater alcohol consumption, lower physical activity, shorter and less consistent sleep, and greater behavioral variability (Ps < 0.001), with stronger associations for alcohol and sleep compared with HRV. HRV-CV increased with age in males after ∼40 yr and showed a U-shaped pattern in females, declining through midlife and rising after ∼50 yr.

HRV-CV increased with BMI in both sexes (Ps < 0.01). HRV-CV measured during nocturnal sleep can be reliably estimated from at least five nights of data, with higher values associating with less favorable behavioral profiles, older age, and higher BMI. These findings support the use of HRV-CV as a scalable, behavior-sensitive digital biomarker with potential applications in personalized health monitoring and risk stratification.
LINK

 

[Mij]

From one of the authors:

"Most people know HRV — heart rate variability — the beat-to-beat variation in your HR. Generally, higher = better autonomic function.

But we looked at the coefficient of variation (HRV-CV). That's how much your nightly HRV bounces around across the week.
Think of it as the consistency of your HRV. A high HRV-CV means your nervous system is less stable night to night — and that turns out to be a meaningful signal."

 

[MrMagoo]

Not sure how I feel about this.
I know HRV variation is part of the Visible monitoring and daily “readiness” score, and that high HRV is a good thing.

This study sounds like it was done on the general population and low HRV is associated with “less favourable behaviour profiles” alcohol use, low physical activity, high BMI.
It might be my medical trauma speaking but I can see this being another stick to beat us with, like exercise.

 

[Jonathan Edwards]

I can see this being another stick to beat us with,

That's how it seemed to me, even if it seemed that those using it were of the non-stick sort.

 

[rvallee]

So it's influenced by a lot of unrelated things, all of which can be independently assessed better by other means. So it adds no meaningful information. It's odd that they didn't look at infections, which WHOOP does record, and for me had the biggest swings. It's weird as they do note that illness influences HRV in the text, but did not use the data.

So, interesting, but not useful. Or about as useful as a weather rock. Rock wet, rain or maybe nearby water gun fight. Rock moving, wind or earthquake. Rock gone, either high wind, thief, saboteur, or string broke, maybe the passage of time, or it was made of a weak fibre for the rock's mass. Or many other possible explanations that can be far more accurately assessed by other means.