Apple does respiration correctly?
Earlier this week Apple announced some of the new health features we will see in watchOS 8 when it is released in August/September. One of the new features of the native Sleep App is that it records the respiratory rate on the Apple Watch whilst sleeping.
In itself, this appeared to be singularly UNinteresting. After all, many of Apple’s competitors already produce respiratory rate data along with other health-focused metrics including HRrest, HRV/SDNN/rMSSD and SpO2. I believe I’m right in saying that subtle variations in optical HR recording are used to determine when a breath is taken. So I initially imagined this was yet another case of Apple coming along a little bit late to the party, perhaps with a slightly better sensor (ie the AW6 HR sensor) and then, as a result, producing slightly more reliable data.
Nope.
The interesting news here is that Apple has taken a different approach as their press release pointed out
Today, Apple Watch helps users meet their sleep goals by establishing a pre-bedtime routine, and tracks metrics like time asleep, heart rate, and blood oxygen. watchOS 8 gives users even more insight into their overall wellness by tracking sleeping respiratory rate, which is the number of breaths per minute. To do this, Apple Watch uses the built-in accelerometer to measure respiratory rate while sleeping…
After a little digging around it seems that the new sensor readings are taken at 50Hz ie fifty times a second and that these readings make assessments of tiny bodily movements that can be quite accurately correlated with the true respiratory rate which is easily validated by independent sensors.
Q: So why is this better?
A: try this
- individual readings may or may not be more accurate than those based on HR (TBC), but there should be a MUCH greater number of these readings and I would guess that overall accuracy is improved because of a greater sample size
- Apple can then give an almost constant nightly feed of respiration data. ie FAR more data points for app developers to work with. This may QUICKLY allow the detection of some unwanted sleep events and triggering exception conditions/alerts.
- Apple currently only takes 2 or 3 nightly HRV readings (unchanged in watchOS 8 so far) and the infrequency of this data limits its usefulness. The respiration data is more frequent and potentially more useful as a result.
Implications
The two main thrusts for this data are towards early onset detection of illness (covid) and Sleep Apnea events. So these are HEALTH-focussed features rather than fitness focussed.
The trick, I believe, is going to come in making a valid fusion of the respiratory data with other health data sets. So an algorithm for detecting upcoming illness might look something like this – increased respiration rate above the 30-day average, SpO2 falls 1%age point below normal and HRrest rises by 5bpm independently of workout stress….something like that. ie not straightforward.
Fitness Take Out
Sadly there is no fitness and athletic take out here. Other than it’s always good to know if you are going to get ill. To get more useful sporting recovery metrics from the Apple Watch we still need MORE automatic nightly recording points please Mr Cook (Apple CEO).
Science link: ncbi.nlm.nih.gov/pmc/articles/PMC7471298/#CR1 “There is substantial evidence that deviations of the respiratory rate from its normal behaviour can be used as a predictor of clinically relevant and potentially fatal events and conditions (see, e.g., the very recent review paper by Liu et al.1 and references therein)” and ncbi.nlm.nih.gov/pmc/articles/PMC7471298/, shows the validity of Apple’s approach.
Thank you: DanielH & MarkusE for the heads up and pointers.
Recommended Reading: Apple Watch 6 Longterm Review
Reader-Powered Content
This content is not sponsored. It’s mostly me behind the labour of love which is this site and I appreciate everyone who follows, subscribes or Buys Me A Coffee ❤️ Alternatively please buy the reviewed product from my partners. Thank you! FTC: Affiliate Disclosure: Links pay commission. As an Amazon Associate, I earn from qualifying purchases.
I have an EmFit QS and it supplies respiratory rate, I wonder if there’s any real difference between measurements via accelerometer and optical. Seems like it’s probably an “easy” number to get either way and I’d be surprised to see much of a difference between the two, will be interesting to see.
That said – I don’t find it all that useful. Almost every day I check my recovery (HRV) and overall heart rate. It usually tracks pretty well with what I expect (e.g. overeating late, being sick, heavy training, etc.. I can usually see clear, direct correlation).
Looking at my respiration rates, it’s almost always 18-19 avg. Sometimes it’s lower, sometimes if I’m sick or something it goes higher, up to 21. Not sure what I do with that, lol. I do use HRV to determine how hard I train.
Final thought – I wish there was some standard ecosystem like Strava where you could upload sleep data and it would tie it all together with everything else and present more useful broad analysis.
good point.
EMFIT uses a 3rd technology…BCG. which i guess is similar to the accelerometer. I imagine it has the easiest job of the 3.
hrv is hard to decipher the information in the wave
and on the wrist there are so many movements going on. that said, at night when motionless if must be easy to detect breathing on the wrist but I’m less sure of the situation in more actives phases of sleep…it feels like it must be harder then.
yeah i think respiration is difficult to know what to do with on its own. hence why i flagged up the need to fuse the algorithm with other sensor data.
not sure i quite see what you are looking for with your final thought over and above what emfit, for example, gives. many of the health platforms can be syncd…look at fitnesssyncer and see if that answers your question. ultimately google fit and apple health aim to be repositories for health data….it’s then up to apps to interrogate it for what they want