Related
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Sleep data comparison between Garmin 945 / Oura v2 / Withings Sleep mat

dpupkov
over 3 years ago

Finally decided to download data from my Garmin 945, Oura ring v2 and Withing Sleep mat and compare sleep metrics for several months worth of nights. Seems (at least for me, n=1) that bedtime and awake time correlates good between devices, while other metrics like sleep stage duration have weak correlation.

Full post at Medium if one is interrested - https://quantify.d.pupkov.com/sleep-metrics-comparison-ea91c6456734?source=friends_link&sk=a23060f76d084a1600691aee6d444f1a

Couple of charts from the post:

As a note - I didn't include HR and HRV data comparison in this post. Based on my experience HR/HRV data between Oura/Garmin/Withings have much better correlation than sleep data.

  • 0 over 3 years ago

    Maybe to complement your observations, I did something similar in the past. Result was that Fitbit and Withings were somewhere midfield, Garmin before they changed their sleep algorithm to their own proprietary one as well, but after they became really bad. This only changed recently on the FR945 when they reverted back to Firstbeat sleep algorithm, this improved Garmin's scoring a lot.

    Oura v2 was above average but only with their latest Gen3 rings this massively changed to the better. I'd say, the latest Gen3 from oura comes as close to a sleep lab as any existing consumer device ever will get. The rest is more or less gimmicks with some sort of correlation.

  • 0 over 3 years ago in reply to HUMODDA

    Thanks

    I agree that original Garmin sleep algorithms were more of random numbers (including bedtime / awake time). That's why I included in my comparison only results after my watch updated to new (FirstBeat) algorithms.

    For Oura, to be honest I don't want to spend another 300$ (+subscription) on a new v3 ring; especially as my old v2 ring is just one year old. I do not see any value in such upgrade.

  • 0 over 3 years ago

    If I may, I'd recommend comparing not consumer products with each other, but to compare it with a baseline provided by (medical equipment) PSG data.

    As it turns out, Garmin et al are getting better at detecting sleep start/stop events, but when it comes to things like REM, their numbers seem a little like reading tea leaves in their data quality.

    Thank you for the interesting data analysis tho! Definitely a fun read. :-)

  • 0 over 3 years ago

    Yeah I wear a 945lte and a 745 together to bed and they both produce different sleep scores so even the same Garmin technique can't correlate 

  • 0 over 3 years ago in reply to 9386496

    Your wear both on the same arm?

  • 0 over 3 years ago

    Interesting but is there a 'proven golden standard'? I mean, is one of these devices the correct one or accepted as the most correct one?

    Other then that I use all these things as I use all of them, general guidelines... Sleep score tends to kinda replicate my feeling. But all the specific details are just, laaarge grain of salt. 

    Same with VO2Max, guidelines... When using the same device at least you can check out trends (VO2Max increasing actually means things get better generally... even though the exact values might be off).

    But interesting and nice to see, would be nice to be able to check them against that golden standard if any.

  • 0 over 3 years ago in reply to MOTIVIO

    During the day I wear Oura either on the left or right hand.

    During the night I almost always wear both on the same (left) arm, and Oura on the (left) index finger. Some nights I may forgot to move Oura from right to left night; number of such datapoints is by my estimation quite small - less than 1-3%.

  • 0 over 3 years ago in reply to FlipStone

    There is actually a 'proven golden standard', which is known as polysomnography (PSG), and it records an actual brain waves. Such labratories are quite not a rocket since, and anyone could go to a sleep labratory, and sleep over a night with multiple sensors on the head, and get a minute-by-minute result.

    Every fitness / sleep tracking vendor (Withings, Fitbit, Garmin, Oura, etc) claim they are indeed validating (and actually train their datasets) against data from PSG.