Vivoactive Respiration Rate - So inaccurate it cant be used

I own the Garmin Fenix 6, and the behavior of the respiratory rate feature is similar. My brpm is typically 6-8 while stationary. The watch reads 14-18. I'm an engineer, and I've been skeptical about how a watch can accurately determine respiration rate. My thoughts (not knowing anything about biometrics or possible indications):

1. Motion: the watch is equipped with an accelerometer and/or gyro to detect motion (e.g. uses a wrist flip to activate the watch light in low light for better face visibility). Perhaps it could filter on subtle, low frequency motion to try and detect motion associated with breathing. Garmin states that if you trap your arm under a pillow or the like that measurements could be less accurate. Perhaps this is why (motion would be isolated)?

2. Heart Rate: I'm very skeptical of this one. Heart rate should be very stable at a given level of activity. Variations about that rate due to inhalation or exhalation?  

3. O2 Sat Variation: I'm not clear on how the watch detects O2 saturation in the first place (perhaps optically, through color of blood vessels at the wrist?). I could imagine O2 saturation varying as a function of influx of oxygen into the bloodstream from the lungs. The period of variation might correlate with respiratory rate. Still, how accurate are O2 saturation measurements?  Respiration rate would only be as good as that estimate.

4. Other?  Perhaps electrical state or some other indication?  

My sleep detection function hasn't worked for a week. Today was the first day, and it seemed (finally) to accurately determine sleep/wake times (whereas before, the watch was wildly off - told me I'd been sleeping in the middle of the day, and ignored 6-8 hours of sleep at night). Perhaps adaptation? I'll keep watch (pun intended) to see if it continues to work.

Regarding adaptation, the watch (or offboard software) would need to apply a learning algorithm of sorts to "adapt through continual wearing". Supervised machine learning algorithms depend on feedback through "training" - telling the algorithm what classifies as, for example, sleep vs. waking hours.  I've been entering sleep times every day for six days which, combined with the recorded data over those times, amounts to ML training data. But there's no way to provide training data for respiration (you'd need to tell the watch when you begin exhaling over many breaths, for example).

Anyway, I guess I'm glad I'm not the only one seeing this problem, but also wished it would just work. Garmin?  Are you listening?

Pretty sure they use Firstbeat Analytics to develop or assist them on their method and algorithm for determining breaths per min from heart rate. Look them up for explanation on how they work it out. Looks proprietary so probably licencing and the such involved.

And to your last point, yes I'd rather they just got it sorted. It's either possible with the hardware on the watch or it's not. If it's possible then fix it and get it working correctly, or if it's not, cool, mistakes happen, offer refunds all round and move on. Consumer Laws are pretty simple in these cases, if it doesn't perform how it advertises itself to perform, then exchange or refund, your choice.

So if you want it fixed, email support and request it to be fixed, at least that gets your problem logged. I don't think these companies read these forums hardly at all or you would think there would be an official reply or something. I've got a request in from December which has gone back and forth a couple of times so probably due to ask them for an update. If they can't get it right then will at least be requesting the option for a refund. Which would be a shame as I really like the watch.

8 months later - the same thing. The respiration rate widget just doesn't work right. When you are in an activity like yoga or breathing exercise, most of the time it shows accurate breathing rate, but outside of activity - doesn't work at all.

Garmin is using Firstbeat's algorithms for this and these algorithms use HRV to calculate the breathing rate, just like the stress widget. In theory, the respiration widget should work in real-time because the watch is measuring HRV in realtime for the stress widget to work, so I don't see why this problem is not fixed yet.

And for some people, this problem may lead to other problems. If someone is having problems with their respiration, but they see their watch showing "normal" values, potentially they may ignore serious health issue. 

Same thing with me. My breath rate is 7, Garmin Vivomove show 14!!

Same thing happens with my Descent Mk2i! I do lots of mediation and breath holds for 3-4 min. When I slow down my breathing to 3-6 brpm the respiratory app actually goes higher (15brpm) I called garmin and they said they would put in a troubleshooting ticket. I paid $1700 for this watch and it’s not even giving me ball park readings 

(my average rhr is 43)

did you return the watch or get it fixed? I just got the descent mk2i and I like you do lots of breathing and meditational exercises including breath holds (on land) and everything you described is exactly what I experienced... bummed.

did they fix it? I have the descent mk2i and same issues! Huge Bummer

To be fair, 'm actually not surprised that the respiratory rate algorithm would cease to function during breath hold exercises. That's an outlier case - the algorithm must assume that you're actually breathing as the basis for detecting variations in data indicating breaths. How can we expect the algorithm to determine respiratory rate if you're holding your breath for > 1 minute?  Let's start by just generating good results for "normal" breathing - the Garmin Fenix 6 does not even do this appropriately.  Mine is regularly off by a factor of 2 (higher than actual).