Hi folks. Here's an email I sent to Zeo support asking them to look into big inaccuracies in Zeo's analysis of my sleep compared to Fitbit and my own observations (links to fitbit and zeo graphs at bottom). Has anyone else seen REM mistaken for wake?
Note: I have observed this both with the table model and with the iPhone version.
I’m trying to really analyze why Zeo is failing, so I bought a Fitbit, which also does sleep analysis, and used them both last night. The two graphs are below.
This seems to be definitive proof that the Zeo is making incorrect conclusions, for four reasons:
1.Fitbit does not show the long periods of wake that the Zeo does, which I don’t remember in any way.
2.The Zeo says that at just after 2am, I quickly went from awake to deep sleep and back to awake again. That’s not even possible.
3.I feel rested after a good night of sleep, and have done so every time the Zeo says I get a really bad night of sleep with being awake for several hours which I don’t remember.
4.I awoke several times last night from a dream, which I remember, but the Zeo only shows it once. This happens a lot with the Zeo, especially in the morning final wake up.
I think what’s happening is that the Zeo is mistaking wake for REM. It would explain all the differences.
Would you please pass this on to your developers? Fitbit is really going to eat your lunch unless you can get this stuff fixed.
And it is possible to fix it: you need to give users a tool to reverse the wake to REM when they know that they were dreaming at certain points; or at least have a slider bar to adjust the graph in that direction. Call it “user normalization”. Given your scientific basis of only 85% accuracy, that would seem to be a good thing to do for the 15% of your audience that is getting bad results, or the 100% of your audience that is getting bad results 15% of the time.