Hacker News Reader

Comment by mg | Hacker News Reader

Is there a good fitness band to measure sleep quality that lets you download the data from a Linux machine, or at least lets you see the data on screen?

With "see the data on screen", I mean at least some quality index, not just the number of hours you slept.

View on HN

I have a Garmin Venu 2 that acts like a thumb drive when plugged in on Linux (and I assume other OSs). The files it stores exercise data as are accessible with a few FOSS tools. You can also put music on it this way. I haven't tried to extract and view sleep data to desktop, but it will show sleep scores and charts on the devices screen. I believe most Garmin behave this way when plugged in, so can be used without cloud/app interaction.

I would be interested to know if others have interesting linux desktop software or workflows for working with this data when it comes off your watch

Contrary to many android wearables, the most recent google pixel watch 2 is reasonably good at sleep tracking according to the quantified scientist: https://www.youtube.com/watch?v=Ef1by8kJfk4&t=1109s

Google partnered with fitbit last year and now sleep data and other biometrics are shared with the fitbit app. The good thing about this is that you can access your data via fitbit's API, there is a python package for this process: https://github.com/orcasgit/python-fitbit. You can read this article if you want more details on how to configure oauth to make this work: https://towardsdatascience.com/using-the-fitbit-web-api-with...

The cons of this setup is that you still need an android phone synched with your watch and is not too privacy friendly, but sure, you could run the code on your linux machine

For me, the Google Pixel 2 is just not a good device. I recently bought one to replace an Amazfit GTR 4. So many wrong things in the Pixel, I went back to the GTR 4. Doesn't track heart rate or 02 during non-sleep periods, for example, and impossible to see individual readings on the phone. Doesn't show enough text of the alerts. ECG readings are awkward and not automatic (though the GTR 4 does not have them, they are so awkward on the Pixel, they might as well not be there - it insisted I switch the device to the other wrist to get a reading).

Oh, and the GTR 4 battery lasts about 10 days for me. The Pixel never lasted more than two.

> Google partnered with fitbit last year and now sleep data and other biometrics are shared with the fitbit app

Google owns Fitbit, and it’s not clear what the future of Fitbit is as a standalone brand or product line

I use Arduino and piezo on the plank under the mattress: https://files.catbox.moe/y6ehk6.gif

But what does that tell you about your sleep cycles and sleep quality? It just shows you moving around in bed.

Sleep cycles are very hard to measure reliably, basically no wrist worn device should be trusted for that.

I find wrist sleep apps accurate enough for showing me trends over time and correlations between things that make sleep worse (e.g. drinking before bed shows different sleeping patterns than not drinking). Interesting and useful data but I agree it's not scientifically precise.

I agree, which is why I think that it would be more interesting, and probably more honest, to show some sort of "sleep quality" index instead.

Isn't "movement in bed" how all sleep monitors work? (aside from the clinical test with electrodes stuck to your head)

I always assumed the main metric used by sleep apps for watches is movement.

>Isn't "movement in bed" how all sleep monitors work?

No.

The best ones monitor your brainwaves, and the best wearables monitor the next best ting, your hear rate and blood oxygen.

You can get knocked out by drugs, alcohol or just die in your sleep from apnea so you'll not move around in bed and your bed motion sensors are useless.

You need to tap into the body's vital signs to know your sleep stages accurately, especially REM.

For anyone interested in how Apple Watch measures sleep stages, here's a deep dive that goes into details about how the accelerator is used to estimate respiration rate, etc in order to classify sleep stages

https://www.apple.com/healthcare/docs/site/Estimating_Sleep_...

I use Withings. Both their watches and their Sleep tracking mat can do sleep quality tracking. I use the latter at home, or their watch when I travel. So I have data no matter where I am.

They have an export/takeout form in the app, which will create an archive with all your info and send you an email, and they provide instructions to parse it (e.g. in Excel). They also have an API, but it requires setting up a public OAuth endpoint for authentication (for Web apps) or pre-approval (for Mobile apps).

Garmin watches expose their data as a mountable filesystem. They're generally very good quality-wise as well.

I've been looking for this too - high quality sleep tracker that doesnt connect to the internet that produces downloadable metrics. I dont think it exists.

If gadgetbridge wrote an oura ring connector id buy one in a heartbeat. I had a mi 5 and the data quality was so crap.

Amazfit watch gives a "score" that can be seen on device screen, in addition to the total hours and and breakdown per awake / REM / deep / light time.

Yes, many activity trackers do this.

It is interesting to wear 2 of the same trackers, one in each wrist, and see if they are close to agreeing.

I do that with 2 Mi Bands.

The variance of a single point estimate is infinite.

Reply on news.ycombinator.com