Fellas, I am looking for some information about the accuracy of the (i) sleep cycles and (ii) sleep stages advertised in Garmin's sleep reports, specifically with respect to REM sleep and deep sleep. Not even that, I'll settle for just some information about how they actually do it. A week into this research got me absolutely nothing. Zero.
=> Yes, I know Garmin guesses, and I have read what Garmin said on their various websites which is essentially... nothing of substance. They collect "data" (which I presume are limited to HR, HRV, movements and... that's it I guess ?), inject these into an algo and voila.
I fail to see how this can remotely tell Garmin how much time I spent in REM and deep sleeps, and how many cycles were repeated during the night.
One thing that I can observe works very well in my case, is sleep detection*. This works very well at the moment. I have a very well oiled sleeping routine, perhaps this helps, I know when I fall asleep and Garmin will detect that with 10' accuracy. Kudos for that, and I hope further updates won't f*ck this up.
Now back to REM sleep and deep sleep stages. A number of things make me terribly suspicous if Garmin is actually detecting anything, for example: there is scientific proof that alcohol inhibits the production of both deep sleep and REM sleep and at best, produces a state of uncoinciousness, not sleep. By extension the brain is inhibited from entering into both deep and REM sleeps**. Yet Garmin advertises me healthy levels of deep and REM sleeps when intoxicated***. A few weeks ago I went to bed drunk as a skunk after a wine tasting, Garmin's report told me the following morning with a straight face that I had over 2 hours of both deep and REM sleeps 
. Even more than I usually have. This is, of course, totally impossible.
In any case, does anyone here have more - substantive - information about what it is that Garmin is doing ? I am probably ok ignoring sleep reports after a night of drinking, as these have become quite rare now, and focus on the "normal" days. I presume Garmin's algos are considering this as a pre-requisite anyway.
* limited to detecting when I fall asleep only. I always end sleep mode manually when I wake up in the morning.
** for anyone interested in sleep, may I suggest Matthew Walker's book "Why we sleep". It's an absolutely masterpiece and explains all this in full details, with EEG and MRI to support his unequivocal conclusions.
*** my sleep scores will be very low because my HRV will be destroyed, but that isn't the point. I am looking at sleep stages/cycles only here.
