Fitbit registering sleep while the user is awake

[diptopol]

We have encountered instances where Fitbit is incorrectly detecting sleep when the user is actually awake.

In one case involving a Pixel 3 watch, Fitbit recorded a total sleep duration of 21 hours and 7 minutes, whereas the actual sleep time was approximately 3 hours and 23 minutes. Notably, the user was not wearing the watch through the whole period that sleep was logged.

In another case with a Charge 6 device, Fitbit recorded sleep even though the user remained awake the entire time.

Would it be possible to investigate these issues further? I can provide additional details if needed.

[LZeeW]

Hi @diptopol    I’m a longtime user, not a developer, but I hope this insight helps you.  My background is physiology with some biomedical engineering tossed in, an added bonus.  Sleep is an EEG diagnosis.  Fitbits use a combination of lack of wrist movement and other parameters to detect sleep.  Many people get sleep detected watching TV or a movie, reading, etc.  With respect to the experience where the user wasn’t wearing the watch, this is what I believe is happening.  The heart rate LED emitter should turn off when the user takes off their device.  I’ve seen situations where the emitter doesn’t turn off.  The light gets reflected back from the band overlapping or off another surface.  An erroneous heart rate gets sensed, the device isn’t moving, so sleep gets detected.

I think it’s important for your users to edit their sleep when these errors happen.

[diptopol]

Hi @LZeeW  Thank you for your elaborate response. Based on what you’ve shared, it does seem like this is an inherent limitation of the current detection method, and as such, we may continue to see occasional inaccuracies like this. Have you seen any official documentation from Fitbit regarding this limitation?

[LZeeW]

@diptopol   This Help article mentions editing start and stop times if the sleep duration seems wrong.  With respect to measuring heart rate when not wearing a device, this is something that I figured out on my own.  Given my background, I've always been interested in how this stuff works.  Users have asked about this heart rate issue for years.  So I did my own experiments.  Violà.  It will record an erroneous heart rate if the emitter doesn't turn off.  There is no official documentation for this.