ResScan: evaluating symptoms &trends in detailed CPAP data

I am new to OSA. 20 days into CPAP, AHI is in the 10-20 range, still fighting nasal congestion before getting to the main objective. I am concerned by the physician’s quick evaluation of just the simple data norms; which technique assumes that the events are evenly spread.
Given some experience in the mathematics of signal processing, I am concerned with 1st and 2nd norms being adequate to describe important higher order phenomena. Like for a runner, how to identify from accelerometer data a leg cramp or cardiac arrhythmia building up; i.e. local rather than global effects in time. I am also concerned with the lack of insight leading to uninformed solutions: like the physician not seeing the plugged up nasal cavity building up at night toward arousal, and prescribing sleeping pills against waking up.

I have enjoyed reading the posts on ResScan data; aware that one flow sensor can't approach the much more expensive sleep study. Still ... I believe there is valuable information in the minute-by-minute or hour-by-hour data, not just tracking the night-by-night averages.
Question to the computer-savvy forum members: have you observed what happens in your detailed graphs as you make therapy changes?

- I suspect that the change in your AHI count is not spread evenly over time, rather bunched in groups. Mine concentrate at the start (ramp up 6-9 over 15 min, even before going to deeper sleep) and at the end, as nasal obstruction builds up. So averaging the whole night would be misleading.
- comparing the shape of one’s breathing curves over time upon a change parameters?
- maybe identifying a sleep stage just from the flow curves
- tracking certain measures not in the data: the total Quality of Sleep for the night, and a cumulative measure of sleep debt; including measures of sleep fragmentation and sleep added without CPAP
- identifying in the detailed data special events: arousals in the middle of the night, changes in body position

Has anybody seen a standardized physician's guideline on optimizing the process? An expert's checklist, written up for students or doctors or sleep technicians new to the field? So, unlike other threads, I don't mean to complain about ResScan vs. Encore discrepancies, or ResScan vs REMbrand PSG results; rather to get more out of the available information within one night. Before too many days accumulate and the machine decides to compress the older detailed data to just averages.

AMUW wrote:I am new to OSA. 20 days into CPAP, my AHI is in the 10-20 range, I am still fighting nasal congestion before getting to the main objective.

Good luck with the cold.

I am concerned by the physician’s quick evaluation of just the simple data norms; which technique assumes that the events are evenly spread.
Given some experience in the mathematics of signal processing, I am concerned with 1st and 2nd norms being adequate to describe important higher order phenomena. Like for a runner, how to identify from accelerometer data a leg cramp or cardiac arrhythmia building up; i.e. local rather than global effects in time. I am also concerned with the lack of insight leading to uninformed solutions: like the physician not seeing the plugged up nasal cavity building up at night toward arousal, and prescribing sleeping pills against waking up.

I have enjoyed reading the posts on ResScan data; aware that one flow sensor can't approach the much more expensive sleep study. Still ... I believe there is valuable information in the minute-by-minute or hour-by-hour data, not just tracking the night-by-night averages.
Question to the computer-savvy forum members: have you observed what happens in your detailed graphs as you make therapy changes?

Well, I'm not one to make lots of changes to my therapy on my own. But with my PA's input, I have had my settings change from straight 9cm, to autopap running 4--20cm for autotitration to autopap running 4--8cm. And since my BiPAP will be set at 8/6, I did change the APAP's range from 4--8 to 7--8cm as a way of helping me get used to falling asleep with the xPAP set at a level higher than 4cm. When changes have been made, I have found that I needed a week (or more) of data to really figure out if anything in the data is different. [How I feel while trying to fall asleep or stay asleep is usually pretty apparent within a day or two though.]

- I suspect that the change in your AHI count is not spread evenly over time, rather bunched in groups. Mine concentrate at the start (ramp up 6-9 over 15 min, even before going to deeper sleep) and at the end, as nasal obstruction builds up. So averaging the whole night would be misleading.

I agree that averaging the whole night doesn't tell the whole story on many (maybe most) nights. I'm 12 weeks or so into therapy. On many nights I don't have enough events to really worry too much about how they clustered. But on some nights, I definitely do see one or two clusters of 4 apneas occurring in a very short time---less than an hour or so. And on other nights I'll have the same number of apneas, but they actually are spread out pretty evenly through the night. And then I've also had some really bad nights where I've had clusters of 10--20+ events occurring in an hour and a half. But since the rest of the night is relatively event free, those 20+ events show up as an AHI in the neighborhood of 3.0 to 3.5.

For me, I've noticed a weak, but positive correlation between having clusters of events and feeling pretty lousy the next day. Now it's important to realize that I don't do the download right after I get up. So my correlation is based on me knowing I've had a total of 5--10 events overnight at the beginning of the day, but feeling worse (sometimes much worse) than other days when I also know I had a total of 5--10 events overnight. It's only when I get around to downloading the data and studying it that I know that I've had a cluster or two of events on the nights before I feel worse during the daytime.

- comparing the shape of one’s breathing curves over time upon a change parameters?
- maybe identifying a sleep stage just from the flow curves

I definitely notice several standard patterns in my flow curves on the S9. But I've not really been able to identify them with any particular change of settings. I also have not been able to correlate the various breathing patterns with a change of sleep position or whether I'm dreaming---after all I'm asleep and I really don't remember when or whether I changed positions during the night. I can usually identify when I did wake up just enough to turn over---but from what old position to what new one? I'm clueless.

Sometimes (not always) there's some evidence in the flow curve that I've either woken up (without remembering it) or maybe gone into REM because my respiratory rate increases slightly and the amplitude of the flow waves increases quite a bit and this happens at a point around 90 minutes into my sleep cycle when it would be likely that I might go into REM. I've even had a couple of nights where this pattern repeats about every 90 minutes. But am I confident that it's REM--no way. It's just an interesting hypothesis for me at times.

What I have noticed that if I look at both the flow curve and the flow limitation curve, there's a definite correlation between some of the shapes I see in the flow curve and readings in the flow limitation curve that indicate a partially obstructed airway. And there's one shape in the flow curve that is often correlated with large bumps on the snore index curve.

- tracking certain measures not in the data: the total Quality of Sleep for the night, and a cumulative measure of sleep debt; including measures of sleep fragmentation and sleep added without CPAP

Keeping a journal for subjective sleep quality might help on tracking whether that's improving with CPAP and how. As for a cumulative measure of sleep debt, I don't think there is on. Sleep fragmentation---other than the most obvious of waking up enough to know you're awake---is going to require EEG data I think.

- identifying in the detailed data special events: arousals in the middle of the night, changes in body position

I think you can learn to identify changes in body position from the flow data. Arousals? I'm much less sure about that. I suppose that if your recovery breath after an apnea, a hypopnea, a RERA or an UARS event has a particular shape that might help. After all the recovery breath is supposed to be tied to why you're waking up, if I recall correctly: Airway collapses, partially collapses, or threatens to collapse; the respiratory event happens; you don't get enough oxygen; brain wakes up just enough (i.e. arousal) to get the body to reopen the airway and take that recovery breath. But respiratory events are only one thing that can lead to arousals and I'm not sure the breath-by-breath data would be enough to pick up other kinds of arousals.

Has anybody seen a standardized physician's guideline on optimizing the process? An expert's checklist, written up for students or doctors or sleep technicians new to the field? So, unlike other threads, I don't mean to complain about ResScan vs. Encore discrepancies, or ResScan vs REMbrand PSG results; rather to get more out of the available information within one night. Before too many days accumulate and the machine decides to compress the older detailed data to just averages.

I haven't seen such a thing. However, at http://www.binarysleep.com (a forum for sleep techs about running and scoring sleep studies), it's clear from the commentary that learning to score full-data PSGs is a very complicated process and that there are two standard ways of scoring hypopneas in use at accredited sleep labs. Likewise the information I can find on the AASM pages about scoring (http://www.aasmnet.org/FAQs.aspx?cid=29) are for scoring PSGs. So I'm not sure there's anything out there for trying to score off nothing but the data generated by a Resmed S9 or a PR System One or other data-capable CPAP/APAP machine.

http://www.devilbissclinicaleducation.com/ondemand

Prepared by Devilbliss, but also refers to other cpap machines.