(03-26-2016, 03:14 PM)0rangebear Wrote: After making offset adjustments to the clock error in the software. I t compare my experimental flagged events to the Flow Rate and Flow limitations Data sets to identify the sleep sessions with the most SWJ. I exported the Data for those nights vis the CSV wizard in sleepyhead and separated our the SWJ form the real events in a spreadsheet / scatted diagram.
It is unfortunate that neither the sleep study folks or the CPAP manufactures have attempted to refine their method to do this. Since insomnia and SWJ clearly skew the numbers.
The problem is that a CPAP has to earn FDA approval, and that approval is necessarily based on its efficacy in treating
sleep disordered breathing, and CPAP manufacturers can demonstrate this without needing to program the machine to try to distinguish wake from sleep. Moreover, there is no way to program the machine to
reliably distinguish between
normal wake breathing patterns and
sleep disordered breathing patterns because there is no way for the machine to
reliably determine whether we are awake or asleep; hence a manufacturer would face a long and potentially protracted battle for FDA approval if they claimed that their CPAP could reliably distinguish which "events" it records in the flow rate data are real sleep disordered breathing events and which ones are normal wake breathing artifacts.
The major problem with attempting to use just the flow rate data from a CPAP/APAP to determine wake vs sleep is that normal REM breathing patterns can and often do resemble normal WAKE breathing patterns. In normal non-REM sleep breathing, the respiration rate is exceptionally stable and the sizes of the inhalations and exhalations are also remarkably stable. There's almost no variability from breath to breath. But in both normal REM breathing and normal WAKE breathing, there can be a lot of variability in terms of respiratory rate and the inhalations themselves can vary quite a bit from breath to breath. The basic difference between REM and WAKE breathing being that normal WAKE breathing can (and often does) include a lot of pauses in breathing that meet the 10-second rule for scoring an apnea or series of deep inhalations followed by shallow enough inhalations to trick a machine into scoring a "hypopnea" that's not real.
Reliably determining sleep vs. wake requires both EEG data and some data measuring movement or muscle tone. Home EEG data (such as what the Zeo used to record) can have a real tough time distinguishing WAKE from REM since REM EEG patterns look very much like WAKE EEG patterns. The makers of the Zeo only claimed about 75% accuracy in distinguishing REM from WAKE, and that may have been part of what led to the demise of their product. (As an aside, SH is able to "import" Zeo data and line it up with the CPAP data.) The actimetry sensors used in things like FitBits track wake/restless periods based solely on how much you are moving around. But it's possible to do a lot of moving around in
normal stage 2 sleep for some people so a FitBit set to "sensitive" mode may over estimate the wake time quite a bit. Conversely, setting the FitBit to "normal" mode may over estimate the sleep time if you actually have a real sleep disorder that leads to a large number of arousals and/or microwakes that last less than a couple of minutes.
On a sleep test in a lab, they use all of the EEG data, movement data (in particular eye movement and facial muscle tone data), and respiratory data to determine the sleep stage and, hence whether any breathing "irregularities" represent normal wake breathing or abnormal sleep breathing. In particular, the movement data can be critical in distinguishing REM EEG from WAKE EEG: In REM, all muscle tone except for the eye movements is essentially gone---i.e. our bodies almost act as if they are paralyzed in REM so that we don't start physically acting out our dreams.
The upshot of all this is the following:
Unless we're willing to sleep with most of the apparatus associated with a formal home sleep test every night, there's just not a reliable way to definitively determine sleep stage (including WAKE) from the data that can easily be gathered in our own beds at night.
But it is possible for us (as individuals) to learn how to read our own
particular flow rate data well enough to make some good educated guesses about when we're most likely asleep vs awake. This requires some subjective data on our part ("I know I was pretty restless last night" vs "I can't remember waking up at all during the night") as well as a good understanding of what kinds of pattern(s) are common in our normal sleep breathing. It also requires having a sense of what a normal hypnograph looks like (i.e. REM cycles occurring roughly every 90 minutes or so with short REM cycles at the beginning of the night with longer REM cycles near the end of the night.) But even armed with all this additional data, at times we may still have to "just guess" whether a particularly nasty cluster of events is just SWJ or real sleep disordered breathing.