S9 AHI accuracy?

[Prowest]

Hi folks...noob here...I have been reading about how great everyone’s AHI results have been that made the switch to the S9 unit. I also purchased an S9 AutoSet unit and will get it this week. I bought it mainly to have a unit that is not as loud (my M-Series has a noticeable blowing sound), but I also wanted an APAP to deal with the times I am not myself and am more prone to snoring (e.g. after a couple social drinks on a weekend)

Some of the more interesting discussions have been around whether or not the improve performance people have been indicating is a function of a superior functioning CPAP/APAP in the S9 or some deficiency in the unit’s algorithm to detect events properly.

I do not know if it has been done, but when I get my unit, I am going to run it in CPAP mode at my M-series settings and see if I get a similar AHI response to the S9 as I do when I am on the M-Series. For CPAP mode, it should not matter what unit I use since the pressure is constant…right? The only difference will be the detection algorithms between the two units.

Any night I have a couple drinks, I am a snoring machine. I typically see a slightly higher AHI results and (on the M-Series) my VS index hits the 20s. I don’t think that Resmed machines record snoring events so it might be a little difficult to equate the two performances, but I’ll give it my best shot!

[DreamDiver]

A couple of us have been taking the same approach - regarding using the machine in CPAP mode first to compare.
They are different machines with different algorithms, but most people are finding their OSA is responding beautifully.
It seems to detect hypopneas differently than all other machines, so that may affect your AHI.
Good luck!

[drjsparker]

Well, I'm new at this as well, but i can tell you that the S9 does record snoring events (mine is pretty much eliminated).

Make sure to get the ResScan v3.10 (earlier versions aren't compatible w/ the S9)

jp

[rested gal]

I do not know if it has been done, but when I get my unit, I am going to run it in CPAP mode at my M-series settings and see if I get a similar AHI response to the S9 as I do when I am on the M-Series. For CPAP mode, it should not matter what unit I use since the pressure is constant…right? The only difference will be the detection algorithms between the two units.

When you do your experiment, be sure to turn off C-Flex in the M series machine and turn off EPR in the ResMed machine.

Some of the more interesting discussions have been around whether or not the improve performance people have been indicating is a function of a superior functioning CPAP/APAP in the S9 or some deficiency in the unit’s algorithm to detect events properly.

I don't know this for sure...just my opinion... different definitions for flow limitation, hypopnea, etc., wouldn't necessarily mean there's a deficiency to detect events properly. Nor would a lower AHI reported by the S9 necessarily be an indication of improved performance.

I'm not knocking the S9 by any means. I think it does have several improvement features. I just don't think lower AHI reports from the S9 compared to the S8 models means there is improved performance, or a superior functioning CPAP/APAP, or a problem in detecting events "properly."

[dsm]

I do not know if it has been done, but when I get my unit, I am going to run it in CPAP mode at my M-series settings and see if I get a similar AHI response to the S9 as I do when I am on the M-Series. For CPAP mode, it should not matter what unit I use since the pressure is constant…right? The only difference will be the detection algorithms between the two units.

When you do your experiment, be sure to turn off C-Flex in the M series machine and turn off EPR in the ResMed machine.

Some of the more interesting discussions have been around whether or not the improve performance people have been indicating is a function of a superior functioning CPAP/APAP in the S9 or some deficiency in the unit’s algorithm to detect events properly.

I don't know this for sure...just my opinion... different definitions for flow limitation, hypopnea, etc., wouldn't necessarily mean there's a deficiency to detect events properly. Nor would a lower AHI reported by the S9 necessarily be an indication of improved performance.

I'm not knocking the S9 by any means. I think it does have several improvement features. I just don't think lower AHI reports from the S9 compared to the S8 models means there is improved performance, or a superior functioning CPAP/APAP, or a problem in detecting events "properly."

RestedGal
Is it fair to say that after so many years of us seeing 'inflated' HI scores skewing Resmed S7 & S8 results when compared to other brands, that Resmed may have finally leveled the playing field at long last. If yes then it now seems Resmed may be being accused of falsifying their AHI scores because it is hard not to read your & some other comments as implying this (ever so subtley & politely).

Interestingly when their (Resmed's) Ai & HI scores were higher than other brands, no one that I ever saw, accused those other brands of manipulating their scores to look better than Resmeds. After all Resmed introduced the 1st Auto machine (Autoset T) & thus became the target for the competition.

Do you have any specific information that *in any way* shows that Resmed's S9 scores are not now on the same playing field as other brands ?

Cheers

DSM

[dave21]

Is it fair to say that after so many years of us seeing 'inflated' HI scores skewing Resmed S7 & S8 results when compared to other brands, that Resmed may have finally leveled the playing field at long last. If yes then it now seems Resmed may be being accused of falsifying their AHI scores because it is hard not to read your & some other comments as implying this (ever so subtley & politely).

I was only thinking the same thing last night that ResMed HI's tend to be higher than in other machines and it could be that ResMed have come down to the same standard as several other manufacturers.

Interestingly when their (Resmed's) Ai & HI scores were higher than other brands, no one that I ever saw, accused those other brands of manipulating their scores to look better than Resmeds. After all Resmed introduced the 1st Auto machine (Autoset T) & thus became the target for the competition.

I would agree with you here.

On the S9 I'm now getting an average of 1-2 a night which is quite a drop from what I was seeing on the S8.

[ozij]

Edit: http://www.resmed.com/int/assets/docume ... ac_eng.pdf

The newly enhanced, intelligent algorithm built into ResMed’s sleep therapy system delivers more effective, accurate therapy to maintain your comfort. We call it AutoSet, you’ll call it brilliant. This special algorithm maintains the ideal air pressure – automatically adjusting through the night to deliver the optimal therapy. Your sleep is more comfortable than ever.
With ResMed’s AutoSet™ range, you enjoy peace of mind because you know the right therapy is always delivered. So you’ll wake up to life - day after day - year
after year.

dsm, do you agree that in its present s9 manual Resmed is hinting, ever so subtly and politely, that the older, A10 algorithm was unintelligent, less than brilliant, and delivered inaccurate and less effective therapy?

If not, do you agree that one can carry this issue of "subtlety and politeness" a bit too far?

None of us know at this point why ResMed's S9 algorithm records less hypopneas than the A10. We only know it does.

I bet one reason they could afford to change their definition is because of their new ability to distinguish between central apneas and obstructive ones -- which mean they no longer have to be so careful about mistakenly raising pressure in response to central events.

Have ResMed changed their definition of hypopnea? I have no doubt they did.
Did the change result in less hyponeas reported by the new machines? Yes.

Is this related to the effectiveness of therapy given by the machines?
We do not know.
Am I accusing ResMed of tampering with the data? No way.
They made a change.
It may mean therapy is more effective.
It may not mean that.
We do not know at this point.

As for tampering with the data, I have more than once noted that another brand, DeVilbiss's low apnea /high hypopnea reporting propnesities can be changed by simply moving their apnea definition from 90% obstructed to 70% obstructed, at which point a bevy of hypopneas will become apneas. The can be done by anyone who knows how to change settings on the DeVilbiss machines -- and is absolutle irrelevant to the way the machine responds to events. Pure cosmetics, by DeVilbiss.

[KatieW]

Ozij, what you say sounds reasonable to me. Hopefully, more will unfold about the S9 algorithm.

Meanwhile, I'm find the new data available (Flow, Minute Ventilation, Flow Limitation) is so interesting and fun to see. I'm not worried about the changes in the "numbers", because I was getting very good therapy with the S8.

For me, the Climate Control was enough reason buy it, and I am very happy with how well that works.

[jmelby]

I think it is important for all of us who pay attention to the AHI numbers reported by the machines to remember that these are intended for trending and guiding treatment, and may not represent AHI the way it would be calculated in a sleep study.

That said, I have looked at my data from the S9 in close detail... I believe it is accurate, possibly more accurate, in reporting events. Looking at the flow graph, it is easy to recognize a stop in breathing and those are appropriately scored as apneas. I'll admit I don't fully understand the mechanism for distinguishing centrals from obstructive, but I can see that there are NOT events where flow "flatlines" for a time that are not being counted.

I also know that the enhanced autoset algorithm is working as intended--I can clearly see the pressure increases after an obstructive apnea event where the pressure is over 10 (as the A10 algorithm would not do since it could not distinguish centrals). In fact, my pressure went higher last night than it ever had before in response to obstructive apnea events. In this sense, I believe the S9 is giving me more effective treatment.

I don't think it is fair to say that the A10 algorithm is flawed--it would be inappropriate to increase pressure after central apneas, so that was the best compromise they could make... use flow limitation and snoring, but not apnea events themselves when the machine cannot tell the difference between centrals and obstructive. The technology has simply advanced enough now to make that distinction, so an update of the algorithm could be made to take advantage of that additional information.

[ozij]

Very well said, and good points on all counts, "yet another Jeff"

O.

[Uncle_Bob]

Given that the S9 is a complete redesign of the machine and hence therapy isn't it really impossible to compare just the scoring of events with the S8

[DreamDiver]

I agree - they are apples and oranges. The best you can do is start with as similar settings as possible and 'feel' your way into making the new machine work for you via self-titration.

[dsm]

[

<snip>

dsm, do you agree that in its present s9 manual Resmed is hinting, ever so subtly and politely, that the older, A10 algorithm was unintelligent, less than brilliant, and delivered inaccurate and less effective therapy?

DSM: No! (? where did you find that ? - I went right through the S9 clin manual & found nothing that fits your description ?)

If not, do you agree that one can carry this issue of "subtlety and politeness" a bit too far?

None of us know at this point why ResMed's S9 algorithm records less hypopneas than the A10. We only know it does.

DSM: I always understood that it was because Resmed's definition of hypopneas was broader that the others. Am sure the early description of hypopnea from Resmed was not nearly as tight as that from other vendors (will do some more research on this as I thought it was well understood).

I bet one reason they could afford to change their definition is because of their new ability to distinguish between central apneas and obstructive ones -- which mean they no longer have to be so careful about mistakenly raising pressure in response to central events.

DSM: Maybe ? but as I said above, the original Resmed def of a hypopnea was much broader - remember the original A10 goes back to the 1990s.

Have ResMed changed their definition of hypopnea? I have no doubt they did.
Did the change result in less hyponeas reported by the new machines? Yes.

DSM: Yup !

Is this related to the effectiveness of therapy given by the machines?
We do not know.
Am I accusing ResMed of tampering with the data? No way.
DSM: My original post wasn't to you so I certainly didn't accuse you. Are you accusing yourself ?

They made a change.
It may mean therapy is more effective.
It may not mean that.
We do not know at this point.

DSM: I think some of us 'believe' there is a difference

As for tampering with the data, I have more than once noted that another brand, DeVilbiss's low apnea /high hypopnea reporting propnesities can be changed by simply moving their apnea definition from 90% obstructed to 70% obstructed, at which point a bevy of hypopneas will become apneas. The can be done by anyone who knows how to change settings on the DeVilbiss machines -- and is absolutle irrelevant to the way the machine responds to events. Pure cosmetics, by DeVilbiss.

DSM: Yes - a very valid point!.

[GaryG]

Given that the S9 is a complete redesign of the machine and hence therapy isn't it really impossible to compare just the scoring of events with the S8

The heck with the numbers, I'm interested in how the S9ers out there feel. Do you feel any different than you did when you were using the S8?

[jnk]

. . . The heck with the numbers, I'm interested in how the S9ers out there feel. . . .

I agree that how we feel is important. But it is good to have some data to get the pressure optimal by using that data for trending.

If a machine tells you your breathing is perfect (0.0 events), it may be (1) that your breathing was absolutely perfect in every respect that night or (2) that the machine has failed to give you any trending data you can use for tweaking pressure. The cynic in me chooses number 2.