CPAPtalk.com

My apologies to anyone who doesn't care for the technical discussions. And my gratitude to those who are patient enough to tolerate it.

-SWS wrote:

Muffy wrote:since the Psearch patent suggests that hypoventilation... will elicit the same behavior as Variable Breathing:

That control layer running Psearch happens to be lower priority than both the VB and hypoventilation detection circuits. So if criteria for either VB or hypoventilation surfaces during a Psearch, then algorithmic control reverts to the highest-priority control layer for which criteria matches.

A comment about the general flow, for those who are trying to follow along. When we look at the flowchart above, we can see a top-down-approach regarding priority. The higher decision branches in that flow chart represent higher-priority APAP routines.

The only difference between that far left vertical branch-pair and the far right branch-pair is that the far left branch-pair is allowed to perform a probing type search for a new optimum pressure. According to the results of an iteration counter or scheduling type criteria, that probing search is either allowed by executing the left branch-pair, or disallowed by executing the right branch-pair instead. Higher-priority A, H, and VB are essentially serviced the same way in either vertical branch-pair. You can almost think of that probing type search in the left branch-pair as something the APAP does during idle time---when higher-priority apnea, hypoventilation, or varbiable breathing do not happen to be afoot.


* * * *

One essential rule for a priority-based algorithm like we see above is that whatever the currently detected SDB event happens to be (A, H, or VB), then the algorithm will temporarily "block off" or close down all lower branches of the flow chart. However, the algorithm will continue to service the higher branches----for the sake of spotting higher-priority criteria. Thus, the search for Popt will not run if any higher-priority decision branches need to service more important SDB events.

Another essential rule, to maintain those priority-based tiers, is that lower-stage subroutines are never allowed to run full course or come to closure without repeatedly checking for those higher-priority criteria matches every step of the way. So when the VB control layer is executing, for instance, only a logical subset or portion of that VB control layer's machine instructions will be allowed to run----before one of many upward-priority criteria checks are made.


* * * *

Plugging those essential rules into a VB-positive detection or case:

1) a logical grouping of VB-related machine instructions are executed,
2) the machine looks for apneas, and services any that are found,
3) the machine looks for hypoventilation, and services any that are found,
4) the next logical grouping of VB-related machine instructions are executed,
5) the machine looks for apneas, and services any that are found,
6) the machine looks for hypoventilation, and services any that are found,
7) the next logical grouping of VB-related machine instructions are executed, etc., etc.

So above, we can see that if no apneas or hypoventilation occur, the VB control layer simply continues to treat VB according to design. The lower-priority Popt search will never be allowed, until there are no signs of higher-priority A, H, or VB.

Similarly, we can see that if an apnea is suddenly detected, that highest-most priority apnea-servicing branch has full control until the apnea is gone. If hypoventilation is suddenly detected in the absence of a frank apnea, the A and H branches are the only two branches that are executed until there are no signs of H and no signs of A.



* * * *

Thanks, Muffy, for posting the flow chart.

-SWS wrote: . . . My apologies to anyone who doesn't care for the technical discussions. And my gratitude to those who are patient enough to tolerate it. . . .

-SWS:

I love the technical discussions, myself. I just wish I was picking more of this stuff up quicker.

If you don't mind my asking, PLEASE keep posting the technical stuff!

And thanks for the help in following it.

jeff

Thanks for the kind comment, jnk. I also enjoy the technical discussions.

This is admittedly a tough exercise in "collective head scratching".

I mean... it's not like working crossword puzzles down at the beach. By comparison, that's a walk in the park... albeit a pretty wet and sandy park.

-SWS wrote:Thanks for the kind comment, jnk. I also enjoy the technical discussions.

This is admittedly a tough exercise in "collective head scratching".

I mean... it's not like working crossword puzzles down at the beach. By comparison, that's a walk in the park... albeit a pretty wet and sandy park.

I usually scratch my head and do crosswords while tapping out an algorithm beat with my pretzel-shaped pen.

Now, while I think "hypoventilation" is simply talking about "hypopneas", since this algorithm is also looking for flattening characteristics, it would seem that there would still be a few "rounded" waveforms left over that still end up stalling the algorithm and go unreported (I imagine these would be central hypopneas).

I have a hypothetical question. If Encore 2.1 files can't be read like Encore 1.x files (using like a html reader and/or Excel) what language do you suppose they'd use?

Muffy

Might try a portable database browser, sparsely hiding among this list:
http://download.cnet.com/1770-20_4-0.ht ... &tag=ltcol

This one's free (haven't tried it):
http://download.cnet.com/Database-Brows ... 15507.html


There's always the possibility Respironics went with a proprietary data structure---perhaps even as an interdiction measure in part.

Thanks very much, -SWS. That (hypothetically) opened the file right up.

I really hate to be a pest, but does anyone know of a good Ancient Phoenician-to-English Converter?

Which reminds me of a funny story...

An Ancient Phoenician walks into a bar and says



LOL!! GET IT?



Boy, those Phoenicians sure have a great sense of humor.

Another observation on human nature...

Did you ever notice when a guy gets a new anything, the first thing he wants to do is take it apart?

Gets kinda messy when you get a new girlfriend.

The MLB season can't be over fast enough (for me, anyway).

What a nightmare.

Looking forward to:



Muffy

Those silly Phoenicians have such a wonderful way with words and jokes. Last I heard they were running a real fancy bed-and-breakfast way out in Scottsdale:
http://www.starwoodhotels.com/luxury/pr ... ertyID=103

Supposedly the place is decked out with all the amenities a traveler might imagine---including complementary copies of "Fun With Phonics the Phoenician Way" workbooks on nearly every coffee table.


_______________________________________________________________________________________________________________________



Since we got so sidetracked, as we often do around here, below is twokatmew's original post:

twokatmew wrote:As a new PAPper and ResMed user, I know that medical professionals claim AHI < 5.0 is "normal." Still most of what I initially read on this forum is about Respironics users seeking (and often getting) AHIs of 1.5 or less. Eventually I found posts by Rested Gal about ResMed machines over-scoring hypopneas and her reasoning for cutting the HI in half. (Velbor has weighed in on this interesting topic also.) Even so, my numbers have been consistently depressing. In recent polls and threads, I'm starting to read about ResMed users "doing just fine" (or not!) with significantly higher numbers. So I decided to start a new POLL/thread to hopefully encourage ResMed newbies (or oldsters) who may be struggling with higher AHIs.

I know there are ResMed users with near-Respironics AHIs, but I'm also realizing there are probably more ResMed users with significantly higher numbers, and this can be OK.

Anyway, off we go....

-SMS, thanks for restating the purpose of the original post -- and especially for explaining the technical points of some responses. It's still a struggle for my sleep-deprived pea brain, but I think I'm understanding a bit more. No matter that we've gone a bit off the original topic. It's turned out to be an interesting thread!

Oh wow, Muffy brings back memories of OU/UT weekends back in my college days. That was a whole bunch of drunk folks.

As the others have mentioned, I appreciate the technical posts by Muffy and -SWS and the others. And both Muffy and -SWS will explain what we don't understand if we ask them to do so. And on top of that, both of them have a great sense of humor.

mar

Muffy wrote:Now, while I think "hypoventilation" is simply talking about "hypopneas", since this algorithm is also looking for flattening characteristics, it would seem that there would still be a few "rounded" waveforms left over that still end up stalling the algorithm and go unreported (I imagine these would be central hypopneas).

Well, I'm not so sure that the Respironics term "hypoventilation" speaks exclusively of what Respironics calls "hypopneas". Here is the definition of "hypoventilation" taken from the above patent description:

hypoventilation is defined as five (5) consecutive breaths with Vm less than 40 percent of the predicted awake supine Vm

Vm refers to "mean inspiratory airflow" here. There are no flattening requirements for those "hypoventilation" events according to Respironics. And unlike Respironics' hypopnea scoring criteria, the comparison is made against predicted awake Vm instead of a typical running baseline established from recent breaths during non-VB sleep.

Additionally, that rather long-duration amplitude-reduction requirement differs from the Respironics definition of hypopnea that Velbor has neatly summarized for us here (thank you, Velbor):
http://velbor.home.comcast.net/~velbor/ ... Defs_3.jpg



So an alternate or perhaps even additional theory regarding why Respironics scores fewer hypopneas than Resmed just may emerge with this hypothetical explanation:
Respironics is considering some of those events to be sustained hypoventilation instead of hypopneas-----whereas Resmed may be scoring them as hypopneas.

I would also wonder whether "hypoventilation" might include "flow limitations" as well as "hypopneas." (ResMed seems to define "flow limitation" strictly in terms of curve flattening; I don't recall how FL per se is defined by Respironics, or whether actual measurement of airflow might be a component.) I have no reason to suggest this other than my observation that FL, in addition to OA and H, appear with Variable Breathing to be in the mutually exclusive set of reporting categories. I was in fact surprised that FL did not have its own decision diamond in the flow diagram, since it is a distinct entity in Encore reports; the answer may be that it is included with hypopnea in the "hypoventilation" diamond.

-SWS wrote:...an alternate theory regarding why Respironics scores fewer hypopneas than Resmed may emerge with this hypothetical explanation:
Respironics is considering some of those events to be sustained hypoventilation instead of hypopneas-----whereas Resmed may be scoring them as hypopneas instead.

This makes sense reading Velbor's posted chart (extracted for those too lazy to click and read the chart):
ResMed

hypopnea...50-75% drop in ventilation...scored if the 8-sec moving average drops below 50%...of the average for 10 consecutive seconds.

Respironics

hypopnea...approximately 50% reduction in airflow...between 10 and 60 seconds compared to average airflow over...several minutes...must see 2 recovery breaths...pressure <8 present in the valley of the hypopnea...terminated by large tidal volumes and reduction in the energy content of the airflow signal...

For for Puritan-Goodknight 420E:

hypopnea...10 seconds...reduction of at least 40%...compared to average of preceding 8 breaths...terminated by 2 consecutive non-hypopneic breaths.

Not sure what Respironics' reduction in the energy content of the airflow signal means but I think -SWS's has zeroed in on the actual cause of the variances between the reporting of hypopneas...Respironics has a more complicated method for scoring hypops as -SWS stated. I did notice that ResMed in Velbor's chart does not include the 2 large-tidal volume 'recovery' breaths; if that is not art of the definition, then there ya are: ResMed scores hypoventilation episodes as hypops.

In my mind, this begs the question, what is the best therapy for hypoventilation? Or is that dependant on the cause? Central obesity as the cause would obviously be losing weight but meanwhile while that is taking place (it ain't gonna happen overnight) what's the best therapy? If caused by muscular or CNS, then the best therapy would be? VPAP because there are other indicators identified in the PSG? Or does this just come down to the CPAP pressure that eliminated the hypops in the PSG? Thanks for additional explanations.

Velbor wrote:I would also wonder whether "hypoventilation" might include "flow limitations" as well as "hypopneas." (ResMed seems to define "flow limitation" strictly in terms of curve flattening; I don't recall how FL per se is defined by Respironics, or whether actual measurement of airflow might be a component.) I have no reason to suggest this other than my observation that FL, in addition to OA and H, appear with Variable Breathing to be in the mutually exclusive set of reporting categories. I was in fact surprised that FL did not have its own decision diamond in the flow diagram, since it is a distinct entity in Encore reports; the answer may be that it is included with hypopnea in the "hypoventilation" diamond.

I don't think Respironics counts waveshape-based FL as if it were a 40% amplitude reduction for 5 consecutive breaths.

Just a reminder that the patent description up for discussion does not come close to representing the complete Respironics design (which is protected by multiple patent descriptions). Some of the other Respironics patent descriptions much better describe the most preemptive control layer dealing with snore, for instance.