[neild]

In my experience, the Apple Watch blood oxygen monitoring was horribly inaccurate. It would report wildly variable results, often telling me that I had a blood oxygen level of 80% (which, if true, would indicate that I should be getting myself to an emergency room ASAP).

Regular pulse oxygen meters are cheap and reliable.

[conradev]

On their best days, they're accurate to within 2-4%. But so many things can trip up the reading, like melanin:

  As a result, for darker-skinned patients, oxygen saturation readings can read as normal when they are, in fact, dangerously low.

https://publichealth.jhu.edu/2024/pulse-oximeters-racial-bia...

When everyone starting looking at every percentage point of their SpO2 during COVID as if it were life or death, the FDA had to remind people of this:

https://www.fda.gov/news-events/fda-brief/fda-brief-fda-warn...

You would be unable to read an accurate pulse oximeter at 80% because you would have lost consciousness. Doctors have to worry about false negatives just as much as false positives with those things.

[mint5]

80% at sea level is very bad. 80% while asleep at 11,500 ft is not unusual. Ref: https://pmc.ncbi.nlm.nih.gov/articles/PMC10157825/

There’s a chart somewhere in there on mean sleep so2 by elevation

[brandonb]

The FDA standard for blood oxygen sensing is within 6% absolute, 95% of the time.

So variability in the sensing is pretty normal, and you want to look at long-term trends rather than individual measurements.

[rafaelmn]

The problem with consumer health sensors is they have both high random error and inconsistent systematic error. When your SPO2 sensor gives you 92% one minute and 98% the next while you're sitting still and it is almost always 2% under, you're not getting "noisy but usable" data - you're getting garbage.

[ayhanfuat]

That caused me nightmares when I was first diagnosed with sleep apnea. I would check my oxygen levels during the sleep to see if my treatment is effective. Even though the CPAP machine would show a few short events Apple Watch would show levels as low as 75%. Thankfully in my next sleep study I learned that my oxygen levels were consistently above 95% and the watch is indeed very unreliable (how snug it is, which direction it is facing etc highly affect the results).

[okrad]

I’ve always felt the sport loops (soft w/ velcro) provide the best contact with wrist while not being too cumbersome. Very easy to tighten just before a workout or loosen before bed. All the while it stays planted on my wrist. Unlock the rubbery band it normally comes with, which is prone to sliding around and less easy to adjust.

Out of curiosity, which band do you use?

[mint5]

It may depend on skin type, body composition and wrist hair - perhaps the validation work used a skewed sample?

I’ve found the sensor to give stable results, with repeated measurements always within 2 percentage points.

And the results give qualitatively very reasonable data when I sleep at high altitude. The readings have a clear dependence on the elevation.

I haven’t cross checked against other meters, but my Apple Watch 9 sensor gives stable and reasonable results that match expected altitude trends. So yeah it may not be tuned to a wide enough variety of wrist types.

[throwaway303293]

In contrast my Garmin and finger pulseox match exactly.

[mauvehaus]

I don't know what Garmin you have, but I'm about half convinced that my Instinct's heart rate measurement is implemented by a PRNG. It's frequently off by 50% from a count/time cross-check.

It does not inspire me to move up their range when this watch eventually dies: if they can't get the basic feature working, I have a hard time seeing how they're going to manage anything trickier.

[exabrial]

Yep, my Garmin also has matched the doctors office instrument to the 1% every time.

[js2]

I've never had any trouble with it on my series 9 (purchased Dec 2023 just before the feature was disabled). It's always closely matched the fingertip meter that I have. Which is to say they both always read >= 95% for the most part.

[shazbotter]

Do you have tattoos? I find most sensors cannot read my vitals through my tattoos.

[llm_nerd]

Indeed, just generally this is a silly feature that was used to sell updated devices, but has almost no value to end users. There is shockingly little diagnostic value of the reading unless you are in such a critical state that you likely want something better than an incredibly unreliable and inaccurate smartwatch feature cram.

For anyone remotely healthy, 100% of the time your real value will be between 95% and 99%, and there is almost no diagnostic value to it. Heart rate is actually interesting and is something you can learn from and work towards. SpO2 is just "eh...neat".

[toast0]

> For anyone remotely healthy, 100% of the time your real value will be between 95% and 99%, and there is almost no diagnostic value to it.

Sure, but if the value is less than 95, that does have diagnostic value (if it's accurate)

[361994752]

as some one whose family passed away due to pneumonia, spo2 is a life saving feature if we had that back then. probably 99.9% of the time spo2 number is good enough. but the value is really about the left 0.1% . of course the false positive rate should be low enough.

[jeffbee]

Wouldn't you already be super dead with a true reading of 80? Or at least unable to cognitively interpret the reading?

[skadamou]

That's definitely a danger zone for healthy people but interestingly enough people with things like COPD may have a blood oxygen level in the 80s and while that is indicative of the disease, they may be totally stable and may not even need oxygen [1].

[1] https://www.drugs.com/medical-answers/normal-oxygen-level-so...

[arjie]

Obviously not. I did the experiment with a finger pulse ox and a Garmin device to check. You just hold your breath. My Apple watch was pretty good at it too. It's very uncomfortable and you'll get visual snow but I'm not dead, super or otherwise. Use your hand to clamp over your mouth and shut your nostrils if you want to try.

[op00to]

I had some momentary readings lower than 80 during a sleep study prior to going on CPAP. I didn't snore, or choke, or anything. Just ... didn't breathe. With CPAP, 98% all the time.