Understand how wearable devices measure HRV (e.g SDNN, RMSSD) to leverage the rich health and fitness information they can provide.
What is HRV?
Heart Rate Variability or HRV is the variability in the inter-beat-interval (IBI) of your heart. It is a hugely important measurement that is widely used across the fitness and health space. Wearable devices have been measuring HRV for quite some time to provide insights to their users.
This variation is controlled by the autonomic nervous system and automatically regulates our heart rate, blood pressure, breathing, and digestion, among other vital tasks. If you’re relaxed, your HRV tends to be higher and if you’re stressed it goes the other way.
For an example of how HRV can be used: see the article “Identifying COVID from HRV”
How is HRV sampled?
HRV is typically measured by wearable devices using either photoplethysmography (PPG) or electrocardiogram ECG/EKG. Most wearable devices use PPG, which relies on illuminating the blood vessels on the surface of the skin and using the reflected light to measure the volume changes during a pulse. ECG, on the other hand, measures the electrical signal associated with a heartbeat. To measure HRV, we must first decide how to sample the heartbeats.
Time domain measures
As HRV is a measure of variability, there are numerous ways to sample a user’s heartbeats to produce meaningful HRV values. The most common and well-tested methods are Root Mean Square of Successive Differences (RMSSD), Standard Deviation of the N-N intervals (SDNN), Number of pairs of intervals that differ by more than 50ms (NN50), and proportion of NN50 over all N-N intervals expressed as a percentage (pNN50).[1]
RMSSD
RMSSD is the most commonly measured form of HRV, it calculates the difference between successive inter-beat-intervals (IBI) in ms, squares these values and takes the root of the mean. The advantage of RMSSD is that the sampling interval can be relatively low and produce a meaningful measurement, with typical measurement intervals of 5 minutes but can be as low as 30 seconds. RMSSD is highly associated with vagal tone, or vagus nerve activity, which in a nutshell is the responsiveness of internal organ activity to stress/stimuli, heartbeats, breathing, etc. It is considered the best measure for short-term variations of HRV, but still a robust measure for longer-term analyses, with typical use cases for tracking stress, sickness, training, and recovery.
Most wearable devices such as Garmin, Fitbit, Oura, Samsung, Eight Sleep, Polar, and Suunto provide HRV data in the form of RMSSD. Terra has integrations with all these devices using our API and you can see the documentation here!
SDNN and SDRR
SDNN is the standard deviation of the IBI intervals measured in ms, NN here means “normal” beats, i.e, removing abnormal or false beats. When assessed over a 24h period, SDNN is the clinical “gold standard” assessing cardiovascular health and in particular, morbidity and risk of cardiac events. You may also see SDRR, which is similar to SDNN, but includes abnormal or false beats. SDNN and SDRR typically require longer sampling times, from 5 minutes at the shortest to the more typical hourly or 24h periods. SDNN is more useful for longer term cardiac health trends and analysis.
Apple presents its HRV data in the form of SDNN. Terra also has integrations with Apple’s HRV-SDNN measurements using our Terra API!
NN50 and pNN50
NN50 is the Number of pairs of successive intervals that differ by greater than 50ms. And pNN50 which is more typically used is the proportion of NN50 over all N-N intervals in a sample, expressed as a percentage. For both measures of NN50 and pNN50 they are highly correlated with RMSSD, and higher values are typically associated with greater cardiovascular health and performance.
Frequency domain measures
Another way of looking at HRV is in the frequency domain, i.e. binning the different N-N intervals based on their frequency in Hz. It is typically separated into two bands, high-frequency HF and low-frequency LF. Activity in each band is represented in signal power. Frequency domain analysis typically requires ECG-based measurements as a continuous time series is required.
HF
The high-frequency band is typically correlated with RMSSD and NN50, lower HF power is associated with stress and anxiety. It is primarily associated with activity in the parasympathetic nervous system.
LF
The low-frequency band is typically associated with the sympathetic nervous system and the HF/LF ratio is often used as a measure of sympathetic-parasympathetic balance in the nervous system, though this is hotly debated.
Importance of measurement period and consistency
It must be emphasized that the same HRV that is measured across different periods such as 30 seconds, 1 min, 5 min, hourly or 24h can vary widely especially depending on the context of the measurement: physical activity, time of day, sleep, rest etc. HRV tracking and drawing insights from HRV is most meaningful when you choose meaningful periods (hourly/24h analysis for basal trends, 1–5min analysis for activity-based responses) and consistency (measure at the same time, in the morning, during an exercise, or overnight during sleep).
See this article on how various devices sample HRV: “Frequency of measurement from wearables:HR and HRV”
Conclusion
A wide range of wearable technologies on the market now measure and present HRV in different ways. With Terra’s API, your business or application can leverage the rich health and performance information that can be described using HRV and other metrics. In addition, by having access to a wide range of wearable integrations through our API, you can be sure that you can find the correct HRV measurement and sampling methods for your application.
Other articles you may be interested in:
- All the wearable devices that measure HRV
- Frequency of HRV measurement from wearables
- Other HRV articles
References
[1] Shaffer F, Ginsberg JP. An Overview of Heart Rate Variability Metrics and Norms. Front Public Health. 2017 Sep 28;5:258. doi: 10.3389/fpubh.2017.00258. PMID: 29034226; PMCID: PMC5624990.