ResMed AirCurve 10 ASV issue

I have an issue with my ResMed AirCurve 10 ASV which began on 1/12/2024. It took a while to conclusively determine this problem. It is a long story, so I will omit the background details. Basically, it goes like this. Beginning on that date, the AirCurve 10 which I had been using for 13 months without issues, suddenly and consistently provided pressure significantly higher than necessary. It was not just that one AirCurve, but two more (one of which was brand new) used later that year that functioned exactly the same way. I have three ResMed S9 VPAP Adapt machines (one of which was brand new in May of last year – yes, I really did acquire a new one – long story) which function exactly as the AirCurve used to function prior to 1/12/2024.

If I were reading the previous paragraph for the first time, I would be highly skeptical regarding its truthfulness. Therefore, I will provide the data to prove my case. Data will be from these seven machines:

Machine Serial number First use Last use
AirCurve 10 ASV (1) 23212972809 4/4/2022 12/8/2022
AirCurve 10 ASV (2) 23212414412 12/9/2022 3/18/2024
AirCurve 10 ASV (3) 23202289977 3/19/2024 4/14/2024
AirCurve 10 ASV (4) 23233753547 4/15/2024 5/1/2024
S9 VPAP Adapt (1) 22131095622 10/7/2013 5/7/2024
S9 VPAP Adapt (2) 22131233829 8/17/2017 5/11/2024
S9 VPAP Adapt (3) 22141242885 5/12/2024 Current

All the data are from SleepyHead reports. Yes, I know this software is old, crude and OSCAR is better. However, it is what I use and it works fine for me.

First, looking at AirCurve(2), the average 95% pressure for nights 1/1 through 1/11/24 was 15.9 and the average peak pressure for those same nights was 21.6. The average 95% and peak pressures for days 1/12 through 1/31/24 were 22.1 and 24.9 respectively. Next is AirCurve(3), for the entire duration of usage, the average pressures were 22.3 and 25 respectively. Next is AirCurve(4) [the new one], for the entire duration of usage, the average pressures were 20.32 and 23.02 respectively. Clearly, all three AirCurve machines are within a consistent statistical range of each other beginning on 1/12/2024. Also, again looking at AirCurve(2) for the entire calendar year of 2023, the 95% and peak pressures were 15.76 and 20.87 respectively, which of course, is very close to what it was for the first 11 nights of January, 2024.

These and other data are shown on the table below (the reason for including the average AHI numbers will be explained later):

95% pressure Peak pressure Average AHI
AirCurve(1) 4/4 through 12/8/22 15.93 20.83 0.474
AirCurve(2) 1/1 through 12/31/23 15.76 20.87 0.362
AirCurve(2) 1/1 through 1/11/24 15.9 21.6 0.448

S9(1) 5/2 through 5/7/24 14.31 17.15 0.08
S9(2) 4/5 through 4/14/24 15.23 19.79 0.294
S9(3) 5/12 through 12/31/24 14.75 18.95 0.15

AirCurve(2) 1/12 through 1/31/24 22.1 24.9 1.52
AirCurve(3) 3/19 through 4/14/24 22.3 25 1.3
AirCurve(4) 4/15 through 5/1/24 20.32 23.02 1.32

The grouping choices are intended to better illustrate this issue.

As can be noted above, all the S9 machines in use after 1/12/24 functioned similar to each other and similar to what the AirCurve machines functioned prior to that date. I also have over five years of data, going back as far as 2013, that show S9(1) and S9(2) were functioning similar to what is shown above.

My DME provider mostly ignored me from May until early last month. It was at that time that a DME technician contacted me and was genuinely interested in solving this problem. After eliminating other possibilities, he too became convinced that the AirCurve had changed. He relayed my information and most of the above data to ResMed and received some feedback from a ResMed technician.

I had originally and erroneously claimed that ResMed changed the AirCurve algorithm on 1/12/2024. ResMed has stated that the algorithm has not changed and is the same for the S9 VPAP Adapt, AirCurve 10 and 11 ASVs. I accept this as a credible claim. However, there is no doubt that the underlying software (i.e., firmware) is different for each model. Each user interface is drastically different and each has some unique options. It is definitely plausible that ResMed could have updated the software via the 4G connection. Also, it would not be the first time that a programmer made what was claimed to be a benign change only to have it affect other functions unexpectedly. Having been a Software Engineer for over 25 years, I have observed this scenario multiple times.

The short story response from the ResMed technician was that the data are incomplete to draw any definite conclusions (my paraphrase). For some reason, both my DME provider and ResMed only has data from January, 2024. The ResMed tech claims the data reported from SleepyHead is not reliable and all data should be analyzed through AirView, the professional software used by ResMed and DME providers. I believe it is also what is used by my sleep doctor. Perhaps there is merit to such a claim, but no comment was given regarding the fact that the SleepyHead report and that from AirView were pretty much in sync for that January. She (i.e., the ResMed tech) has yet respond to my offer to provide the SD cards from the machines still in my possession, namely, AirCurve(4), S9(2) and S9(3).

She also speculated that the higher pressure may be a result of more leakage, which was observed after 1/12/2024. I adamantly disagree. The actual problem is the higher pressure is causing more leakage. While it is true that the machine has to work harder when leaks occur, it does so in order to maintain the required pressure. Just as an HVAC blower has to work harder to respond to duct leakage. The pressure within the duct does not increase, but the power used to maintain that same pressure increases significantly. A user may sense a higher pressure when leaks occur, but what is really happening is higher air velocity due to those leaks. Air is being pumped into the mask at a faster rate in order to offset the leaks and maintain desired pressure. Also, if she were more observant, she would have noticed that there were at least four nights after 1/12/2024 during which there were almost no leaks and the pressures were about the same as the other nights in question.

She also pointed out that there are more apnea events recorded after 1/12/2024. And sure enough, the average AHI for the first 11 days of January was 0.448 and for the rest of the month, it was 1.52. Clearly, that makes her implication credible that the machine is simply responding to a need rather than being faulty (this was actually my original conclusion when this problem first occurred). However, as is the case with all statistics, you can make the data represent different things depending on how it is analyzed and compared. As can be seen on the above table, the average AHI for the three different S9 machines used last year, are closer to the what the AirCurve was in the
early part of January than in the latter part.

Furthermore, the average AHI for the entire year 2023 (i.e., 1/1 through 12/31) for that same AirCurve was 0.362. This is much closer to what I am currently experiencing with the S9 than to the AirCurves after 1/12/2024.

Consequently, it appears that for whatever reason, beginning on 1/12/2024, the AirCurve is perceiving more apnea events than before that date. What has driven this change is puzzling. I have not changed. None of my medical conditions have changed. I have been using the same mask and settings for over five years. Again, the S9 machines currently in use are consistent with what the AirCurve was prior to 1/12/2024.

Perhaps the AirCurve software was tweaked to be more sensitive to apnea events. Maybe for most people this is a good thing. Clearly, for someone like myself who already has very high pressure requirements, this is a disaster. I suspect other people are experiencing this same issue and perhaps for them, whose pressure requirements are not as high, they just don’t notice it since the change is so subtle. Even though it was stated that the algorithm did not change, no comment was given regarding whether or not the software has changed.
Based on the average AHI numbers, one could make the case that even before 1/12/24, the AirCurves were sensing more apnea events than the S9. However, the difference was minimal and had no impact to my sleep or ability to keep the mask from leaking.

To further resist any resemblance of acknowledging an “upgrade” to AirCurve last year, ResMed suggested to check the following:
1. When was the patient’s masks, hose, filters, and humidifier changed?
2. Are they due for replacement supplies?
3. Is the correct mask type chosen such as full-face vs nasal as the mask type helps to calculate for intentional leak?

All masks, filters and humidifier bins are changed as recommended. The same mask type (with some, unrelated exceptions) has been used for over 12 years. Although the S9 and AirCurve uses different hoses, these data differences cannot possibly be driven by a hose issue. I don’t just use one hose, I use five. I have five active hoses at any one time. While one is in use, the other four are hanging up to dry. I used the same five AirCurve hoses before and after 1/12/24. There is absolutely no chance whatsoever that all five hoses would suddenly become defective at the same time.

All of the data overwhelmingly suggest that all the AirCurve machines are functioning drastically differently after 1/12/24. So far, it appears that ResMed is having a “neither confirm nor deny” attitude.

I posted a similar topic (i.e., thread) on ApneaBoard with background and more details. One user posted in late December that he was experiencing the same problem. Unfortunately, I did not see his post for about two weeks. I responded and also sent him a private message. I suppose he gave up since I never received a response. It is hard for me to believe that no one else is experiencing this issue. Again, perhaps those whose pressures are low to begin with are seeing higher pressure, but the difference is so subtle, it is not noticeable.

The ideal solution for me would be for ResMed to provide an AirCurve 10 ASV that functions as my first two such machined did prior to 1/12/24. Meanwhile, I will continue to use my S9 VPAP Adapt.

What would really be helpful to me would be for AirCurve 10 ASV users to check pressure data before and after January, 2024 and if different, share that data. If I am correct that software was updated, it is unlikely that all machines were updated on the same day. If you don’t have access to that data, perhaps your sleep doctor or DME provider does.

I apologize for the way the data are shown. Prior to submitting the topic, everything lined up fine. Then after submission, it all got bunched together.

Terrapin wrote: ↑

Mon Jan 27, 2025 3:09 am

I will provide the data to prove my case.

Yeah, that data really doesn't prove anything.

Sleeprider asked for charts but you said

Any further report of specific data would be redundant.

So he closed the thread.

What would "prove your case" would be breath-by-breath analysis and reverse engineering the machine response. Then you could tell them what the algorithm changes were (seems to me the trigger for a 5 - 7 cmH2O pressure increase would be easy enough to spot).

Terrapin wrote: ↑

Mon Jan 27, 2025 3:11 am

I apologize for the way the data are shown. Prior to submitting the topic, everything lined up fine. Then after submission, it all got bunched together.

Next time take a screenshot of the properly-formatted data and then post the image.

Nocibur wrote: ↑

Mon Jan 27, 2025 12:53 pm

Terrapin wrote: ↑

Mon Jan 27, 2025 3:11 am

I apologize for the way the data are shown. Prior to submitting the topic, everything lined up fine. Then after submission, it all got bunched together.

Next time take a screenshot of the properly-formatted data and then post the image.

The reason for that suggestion is that the forum software messes up tabulated data when it's text. An image will let you present it the way you see it.

I just tried to paste the screenshot, but it didn't work. I can paste it to a Word document, but when I attempt to paste it to this text area, nothing happens.

Okay, now I get it. I think this worked.

Although that wasn't really necessary as forum members can simply "quote" your discombobulated data and it can be viewed (more or less) in its original state:

Much to my surprise, yesterday, my DME provider issued a new AirCurve11 ASV for me to try. I reluctantly agreed only because the rep who arranged for this seems to be the only one who is truly interested in solving my problem.

After using it last night, I was quite surprised at the results:

95% pressure: 15.38
Peak pressure: 24
AHI: 0.46
Mask leakage: virtually none

This is a huge improvement and close to how the AirCurves were functioning prior to 1/12/2024. On the one hand the peak pressure may seem a bit high, however, those peaks (five total) had very short durations and only one of which resulted in mask leakage. If the report had not shown the leakage, I would not have known since it did not wake me up.

Bottom line: This is very encouraging, but of course, just one night. I think I need to continue to use it for at least a week.

So what makes you believe that the latter machine provides the more appropriate therapy vs the former one(s)?

I mean, "different" doesn't necessarily mean "wrong".

Perhaps you've been on suboptimal therapy all this time and the "new" machines are providing the more appropriate therapy.

So again, rather just looking at numbers to base conclusions, one needs to look at exactly why machines respond the way they do.

This is from a brand new out-of-the-box A11 APAP (ASV uses a pile of additional parameters in its algorithm, so this is the tip of the ice cube):

So I'm looking at this and going "Wow, that's pretty aggressive!"

I mean, I'm not really seeing anything that by rule should have been addressed.

In this case, is it good? Bad? Indifferent?

Of note (or coincidence) this area is in what might be a REM and/or supine sleep state, in a person with known REM and supine dependencies.

So does the algorithm know something I don't know?

Need to do a little more fiddlement...

Nocibur wrote: ↑

Sat Feb 15, 2025 4:31 am

I mean, I'm not really seeing anything ...

That said, I should also take a look at the waveform in Oscar, as IIRC the SleepHQ waveforms do not have the same level of detail as Oscar.

Nocibur, you are “barking up the wrong tree”. This problem is clear and straight-forward (albeit with a long and convoluted description). On 1/12/2024, ResMed updated the software on my AirCurve10 (as well as two other AirCurves used since then) rendering it usable for me without drastic and expensive changes to my mask. Most users likely did not even notice since their settings are not as high. Although ResMed has not yet admitted this occurred, they have not denied it either.

You say you don’t believe this is possible? Here is a statement on page 2 of the AirCurve11 User Guide:
Over-the-Air Download of Software Updates. If the device is connected to the cloud, then the ResMed Software on the device will automatically and periodically download updates and upgrades to the ResMed Software on the device.

This statement comes directly from the ResMed web site:
Your ResMed Air11™ device comes ready to receive remote firmware updates, also known as over-the-air (OTA) updates.

Although there is no documentation indicating this applies to the AirCurve10, it has the same wireless capability. It can transmit data and receive changes to settings, so it stands to reason that its software could also be updated.

What is rather interesting is the AirCurve11 seems to be functioning more like the AirCurve10 did prior to 1/12/2024 instead of after that date. The data overwhelming indicates such. If you refuse to see that, there is not much point in discussing it. You have your opinion and I have mine. Clearly, neither one of us will be able to convince the other.

Terrapin wrote: ↑

Sat Feb 15, 2025 7:26 am

Clearly, neither one of us will be able to convince the other.

Yeah I knew that when you refused to submit data ("evidence") to Sleeprider on Apneaboard.

But I like to post images and conspiracy theorists really don't bother me so it's academic anyway.

That said, if you read my last post (or better yet, try to understand it) instead of instantly getting your panties in a bunch you might note that the algorithm of a brand new A11 does appear to be more aggressive than responses previously observed in the algorithm, and might, in fact, lend support to your findings.

Therefore, the sooner you get your fat fucking ass off this thread, the sooner I may continue with my observations and analysis.