Seattle Forrest

I'm not a doctor or sport scientist. I know that people have settled on RMSSD. Google has some info comparing that against SDNN.


As many of you know, HRV can be computed in many different ways, starting from our basic unit of information, the RR intervals (beat to beat differences in instantaneous heart rate). The sports science community through the work of many, including Martin Buchheit, Daniel Plews, Paul Laursen, Andrew Flatt, Martin Esco, Fabio Nakamura and a few others, in the past 10-15 years, settled on rMSSD as the most meaningful and practical feature to use in applied research and real life, when working with athletes.

Heart Rate Variability (HRV) features: can we use SDNN instead of rMSSD? A data-driven perspective on short term variability analysis10/12/2018 Blog post by Marco AltiniOne really good thing about the sports science community, is that we have settled on what feature to use when we talk about heart rate variability (HRV). As many of you know, HRV can be computed in many different ways, starting from our basic unit of information, the RR intervals (beat to beat differences in instantaneous heart rate). The sports science community through the work of many, including Martin Buchheit, Daniel Plews, Paul Laursen, Andrew Flatt, Martin Esco, Fabio Nakamura and a few others, in the past 10-15 years, settled on rMSSD as the most meaningful and practical feature to use in applied research and real life, when working with athletes. Why rMSSD? Well, first of all, most sports scientists are physiologists, they know what they are talking about when considering physiological processes in the human body, and it turns out what also came up from all these studies, is that there is mainly one thing that can be measured using short term HRV features: parasympathetic activity. Without going into another primer on HRV (see this post if you are looking for one), parasympathetic activity represents our body's rest & recovery system, and can be captured in terms of HRV: a stressor might for example induce a physiological response in terms of reduced parasympathetic activity, which translates into lower HRV as the nervous system modulates heart rhythm in response to such stressor. Parasympathetic activity acts quite fast, in the matter of seconds. How do we capture these fast changes? rMSSD, due to how it is computed (just math), captures fast changes in instantaneous heart rate, hence it reflects very well parasympathetic activity. It's also easy to compute and standardized, hence we can be certain we all talk about the same thing, which is a good starting point.

Wonderful, we have a feature that everybody agrees on, and has also a clear link to how physiology works. All problems are solved and we can use HRV4Training or our favorite HRV app to gather data, compare results, and learn a bit more about how our body responds to training and life stress.
Well, not so fast.

​Unfortunately, Apple decided to provide no access to raw PPG data in the Apple Watch and no access to RR intervals either (like a Polar strap would do), as well as not to implement standard communication protocols (e.g. the bluetooth smart heart rate profile). This means we cannot compute rMSSD or any other feature ourselves, as we'd do with another sensor. Additionally, they decided to limit communication between the Watch and any app just via the Health app, and use only one HRV feature in the Health app: SDNN. The reasons are of course unknown and undocumented. There is no access for developers, and no possibility to communicate with anyone involved in this work at Apple. Anyhow, now that we got as far as understanding what is used in Health (SDNN), and we know all the limitations around it, the problem remains, how do we deal with it?

https://www.hrv4training.com/blog/he...ility-analysis