I've started looking at some Emfit QS plots. The first that jumped out was that it thought I was asleep while I was actually watching a movie in bed and moved around quite (I know, not the best sleep hygiene - blame the girlfriend ). Anyway, the Emfit band covers only half the bed and I was roughly in the middle.
This seems to be a data interpretation problem rather than a sensing one, because breathing was clearly detected only after I actually went to sleep around 2:45AM:
Toss & turn detection confirms that:
This false positive is quite common. Here is another Sunday afternoon movie:
Notice how no breathing was detected:
I'll reach out to the Emift CEO with this feedback - maybe they can adjust their sleep phase detection algorithm to consider no breathing = awake.