(04-08-2016 07:59 AM)robysue Wrote: How does the device you are using to track sleep stages work? And what is its accuracy?
REM sleep can be quite difficult to pick up for a number of reasons. The old (defunct) Zeos which used EEGs were only abotu 75% accurate at distinguishing REM according to the fine print stuff that Zeo wrote in their owners guides. Accurate enough for home use, is how they put it.
The device is the ResMed S+. Based on what they have said in the marketing, the device is basically a RADAR unit that uses a frequency at which the clothing, blankets, etc. are transparent, and the body (water mostly) is opaque. It then monitors the movement of the body, I assume by doppler effect. I have not found a technical description of the methodology yet, but I expect it uses a something along the lines of an FFT of the doppler shift rates to create a frequency signature of the movement over a period of time. Then that signature is applied to some form of comparison algorithm (probably an AI Neural net?) that has been taught what various sleep stages "look like".
All that is probably wrong - but... I wrote the software for a similar idea ages ago that was used in a system for cargo ships where the system listened to bearing noise in the drive shaft and predicted when the bear was going to fail, so preventative maintenance could be performed before heading out from a ocean crossing. The system was expanded to "listen" to entire machine shops and make predictions of all the various pieces of equipment, mills, presses, lathes, etc. from the composite noise based on recognizing how each piece of equipments signature sound was changing. Towards the end of my involvement in that product we were able to create usage reports indicating how each piece of equipment was being used in the busy machine shop, correlate that to which machinists were assigned to or using each and determine if someone needed additional training on proper techniques to reduce machine maintenance times.
Anyway, back to S+. If I recall correctly ResMed claims to have tested/trained their system in numerous Sleep Labs during clinical sleep studies and it performed with a reasonable accuracy when compared to the studies results. Since it is not sold as a medical device, "reasonable accuracy" is what ever someone wants it to be.
In my case, I am tracking my CA's and OA's in SH and comparing them to my arousal events reported by the S+. There is an almost 100% relationship between a waking event on the S+ and an apnea event in SH. Of course not all apnea events result in an arousal event, and so there are many OA's and CA's that do not result in my waking up. Also, I can't say how many times the S+ misses me waking, I can only say it is "reasonable" that I would have woken at the reported time because there was a long duration Apnea event (30seconds or more) at the same time the S+ says I woke.
Added to that times that I know I woke because I noted the time on a clock before going back to sleep.
I have not figured out how to determine the validity/accuracy of the light sleep periods reported by the S+.
There is a strong correlation between being woken from deep sleep (according to the S+) and my feeling very "groggy".
On REM stage, in all cases where/when I recall being woken from a dream, the S+ reported I was in REM stage.
I have only been woken a low percentage of the times the S+ reports I am in REM, so it will take a while to draw any conclusions. But it is 100% so far in that specific case. Obviously you dream mostly (not necessarily only) in REM so it is not a perfect qualification.
Also, the S+ may be missing REM states - hard to know what is not there and I don't wake up.
Back to the AHI I am seeing. Yesterday I spent some time adjusting and fiddling and got rid of most mask leaks. Previously the leak rate was running around a constant 20L/min I have it now down to around 4L/min. (sadly shaving my short beard was a major contributor to improving my leak rate - sigh.)
Quote:But leaks would probably not be the cause of a whole string of apneas happening in the first place. Unless the leaks were repeatedly waking you up and the apneas are occuring just as you're trying to get back to sleep. In which case they could be mis-scored sleep onset centrals, which are usually considered normal. Unless they're happening so frequently that can never make it over the hump of sleep transition and actually get t sleep.
This comment I think is right on. What I "think" is happening now, with the leaks reduced, is that I have an arousal event that results in my tossing and turning, which results in my hitting my mask and a major leak occurring. Then I wake enough to adjust the mask and try to go back to sleep. (This awakening is reported by the S+). Then "sleep onset" Centrals occur causing the entire sequence to repeat - often, not always.
So I am going to try going to sleep on my side and not my back to see if that reduces the sleep-onset centrals and focus on not tossing and turning - well, I can try to anyway.
Thank you for your insight, I appreciate it.