(03-26-2016 09:58 PM)Carbon Wrote: I've been using this for 7 days now in the quest to find something that can measure sleep quality given retail EEG's don't exist at the moment.
I offer some comparison vs Sleephead, and also vs a cheapy $80 wi-fi time-stamped intra-red video camera pointed from high above bed head at my legs + Voila for continuous screen capture for last 2 days (I have periodic leg movement). It doesn't take long to review a whole night at x30-60 playback
My overall impression is that the S+ is handy, but gives a surprising negative correlation with Sleepyhead (albeit on a short series versus low AHI's). I can't really guage if it reasonably identifies REM sleep, but it maybe OK for deep sleep (with caveats), and seems to identify "Awake" pretty well, even if I am not aware of short periods of this.
My overall S+ data shows adequate sleep time averaging 7.0 hours, + 50 min awake time. I average 4 awakenings a night (2-8) - I am only aware of 1-2 of these. My "REM sleep" is only 82% of predicted (49-91%) and "Deep sleep" at 91% predicted (71-100%). Overall S+ "Sleep score" 89% (85-97)
My Sleepyhead AHI for the same period is 1.0 (0.5-1.62).
Correlation coefficient AHI and S+ "Sleep score" is MINUS 0.7 (Excel CORREL function). So according to S+, the worse my AHI, the better I sleep!
S+ vs Sleepyhead
The grossest data is accurate - disconnecting/ getting out of bed vs "Awake" -spot on.
Unaware/ unremembered "Awake" periods - correlate with lots of movement on IR camera - probably true
Time to sleep onset pretty good too - S+ vs when Autoset 10 autoramp starts - usually agree within a minute or two
MY AHI is usually under one, so not a lot to look at there, but at least 1/2 my centrals show as a big breath or two before rolling over in bed (confirmed on IR camera.) These tend to correlate with awakenings or "Light sleep" so that looks OK
The rest of my comparison is vs IR camera on my legs
"Deep sleep" looks to be absence of gross movement when asleep, combined with what looks to be fairly steady respiratory rate on Sleepyhead. Fine movements have no effect on "Deep sleep".
Absence of steady breathing while laying still doesn't seem to register as "Deep sleep" (so you can't fool it laying awake)
Possible artefacts in "deep sleep" measurement.
1. wife rolling over close to me and her being in "Light sleep" registers me as being in "Light sleep" despite me not moving/ changing respiration = underestimates my deep sleep.
2. I do have periodic leg movement that occurs in long runs 20-80 minutes (on the camera) with no change in "Deep sleep". Without EEG, impossible to tell if PLM irrelevant, or causing micro-arousals - the latter would mean "Deep sleep" is overestimated.
"REM sleep" - basically little idea. Video looks similar to gross movements of light sleep, maybe without major position changes. Breathing can be choppy or regular. My PLM can occur in "REM sleep". I don't know if I can bring myself to video my eyes when asleep - looking at my own legs is weird enough! The "REM sleep" number seems to have some relation to how fatigued I feel by day, but I did not keep an actual score to document that.
So my impression is the smart alarm seems good (one post up), "Deep sleep" could be OK as long as a partner doesn't get too close (and you don't have PLM), can't comment much on REM.
I'd be great if you can sort out a data-export Frank - eyeballing this stuff is very rough-and-ready!
That is a great amount of info - thank you for posting it. I have 3 days of recording now and I am expecting to get my AirSense 10 Elite next week, so I can start comparing also.
I had not looked at the data feed possibility, but I will start looking into it tomorrow.
My "impression" of the S+ data is that it is fairly accurate when it works - meaning when nothing unusual happens, like my wife rolling into the scene, or my legs don't bounce around. I also question the REM state measure. In my initial sleep study I only had 1 short period of REM, and the S+ is reporting several each night.
I don't use the alarm yet, and I have found if I wake and lay there thinking for a while (15 minutes or so) it does not always register that I am awake.
More later, thanks again.