Comments on Data Please

[rkl122]

.....................
I was reading the bit about Response to OA/Hypopnea events by the Respironics System One RemStar Auto machine in Table 2, and I don't understand all of it.

I almost never say this, but I think this stuff would be easier to understand in equation form.

It says:

"If 2 apneas or 1 apnea/1 hypopnea or 2 hypopneas-increases by 1 and holds for 30 s.

Okay, that first bit makes sense. Two events happen then pressure is increased by 1 cm.

Check.

I don't understand the timing of events after that. Does it mean that another pressure increase can happen after 30s if two more events have occurred at that point? And on and on with a pressure increase as often as each 30s until the limit described in the next statement is reached?

That's the way I read it, yes.

*NRAH logic limits max pressure to 11 or 3 higher than preapnea baseline.
*NRAH = nonresponsive apnea/hypopnea logic

??? Does that mean:

If starting pressure = 4-8, pressure can increase up to 11 cm.
If starting pressure = 8.5-17, pressure can increase by up to 3 cm.
If starting pressure is >17, pressure can increase up to max of 20.

Does that sound correct?

It sounds plausible, but I'm not sure. I assumed that they assume the baseline pressure will always be below 11, so that 11 is max. What you call "starting pressure" is what they call "preapnea baseline," which I take to mean the initial pressure (after any ramping) of the session. The text indicates that there are algorithmic precautions against "over-titration," which may or may not be relevant to how we should interpret the summary. But we shouldn't have to interpret - it should have been expressed more clearly.

If more apneas within 8 min decrease pressure by 2/15 min down to 1 over level that prevents snore then holds pressure for 10 min.


Does that mean "if NO more apneas within 8 min"? I don't understand why the algorithm would want to decrease the pressure if more apneas were happening.

Me either. I think you're right - there's an egregious typo.

Anyway, if more apneas/no more apneas occur within 8 minutes (8 minutes after the last pressure increase?) then decrease the pressure gradually by 2 cm over a 15 minute period until a pressure is reached that = (Psnore + 1) with Psnore being calculated by another bit of code. Hold that pressure for 10 minutes.



Pressure will continue to increase in response to 2 hypopneas."


Only hypopneas? What happend to two apneas or 1 apnea/1 hypopnea?

I'm very confused.

Yes, quite confusing now that I look at it in detail, and there's hardly any clarification in the text either. (Read the "Resperonics paragraphs" in the AutoCPAP section.)
I think much of the confusion is simply due to lazy editing. Although, this may also be an allusion to the fact that a hypopnea is detected from flow data alone, whereas the others depend on flow data plus pulse responses. Again, we shouldn't have to speculate.

BTW, note that some of the key references are to patents. I used to work with patents. There's no guarantee that, just because Resperonics is assigned particular patent technologies (which can include algorithms), those technologies are the ones actually incorporated in their products.

GW, I'm with you, but I suspect there are few of us here interested in the nuts and bolts. Interesting, since in a sense we're betting our lives on these algorithms and the machines that rely on them. OTOH, they are FDA approved devices (someone correct me if that's wrong), so we can assume both the algorithms and the machine specs get the same scrutiny as a new drug application. I'd love to understand better how they work, but gotta confess at moment my focus is getting my AHI down into the safety zone.

Note to self: reserve future thread for penetrating and totally boring examination of CPAP algorithms.

.............. Table 2 summarizes the different event algorithms by manufacturer.

Auto CPAP machine algorithms, etc.

-Ron

[green wings]

Thanks, Ron, for reading all that! I waded through it mostly because I have been running my fixed pressure PR CPAP machine in auto-trial mode recently and haven't liked the way the algorithm behaves at times.

I seem to have more severe apneas during one or two periods each night. I'm not sure if it's during REM sleep or if it's positional. I had no REM sleep at all during my initial sleep study, so that's no help.

What I see is that the PR algorithm increases the pressure nicely when the fireworks start, but then if I don't have another apnea for what I think is around two minutes, the algorithm starts to decrease the pressure. That means apneas start happening again, and it's all just a big mess after that point.

I would like to try a machine with the ResMed algorithm and the ResMed "For Her" algorithm as well. I'm ready for a smart algorithm that adapts itself to my breathing and apnea characteristics.