Dreamstation Central Apnea detection

[sleepyp]

Can anyone comment on the Dreamstation's Central Apnea detection algorithm?

In my sleep study no centrals or obstructive apnea events were detected.
I had an AHI of 29.1 of all hypopneas, with AHI during REM of over 40. The study didn't seem to measures RERAs or Flow Limits. Oxygen desaturations went as low as 86%. I had bad daytime sleepiness and was even falling asleep during my favourite football team's games.

Since starting treatment on a Dreamstation Go 3 months (8-20cmH2O and various pressure setting experiments, tried all Flex settings including off) the APAP has consistently been raising pressure to treat apneas it was detecting.
Pressures were frequently getting to 16cmH20+. It was showing almost all apneas and centrals, the apneas appearing to be what drove the pressure up.

On saturday I received a Resmed Autoset 10.
Since using it (@ 8-20cmH2O, EPR3) the highest pressure has been 10.3, and it sits mainly at 8.3 as an average.
My hypopnea index average has been 0.25 for the three nights.
No obstructive apneas have been detected for three nights, just like my sleep study.
I've been feeling great the last three days, as opposed to how I felt with the Dreamstation Go.
I have been having centrals of around 3 per hour. I think this might be because I'm using EPR at 3 and I'm not used to this level of ventilation, my Tidal Volume has increased quite a bit.

My questions:
1. Is the Resmed better at detecting Centrals? It seems like the Dreamstation was detecting many (treatment emergent in my case) Centrals as Apneas and ramping the pressure. Remember, my sleep study found no centrals or apneas, and neither has the Resmed machine since I've started using it. **
2. Is EPR significantly better than Flex at dealing with RERAs/Hypopneas/Flow Limits? It certainly seems to be the case for me. Why would that be? The Resmed also seems to detect Flow Limits and reacts very quickly, whereas the Dreamstation Go didn't seem even detect Flow Limits.
3. Am I imagining all of this? :-)

** From what I can see the Dreamstation uses a single pulse to determine Apnea vs Central. These pulses are not used for every event and it seems to mark non breathing events without a pressure pulse as Obstructive. It then drives the pressure up causing even more Centrals and (potentially incorrectly detected) Apneas.
The Resmed appears to use some kind of oscillation for every non breathing event, and it marked all of mine as Central and not raised the pressure.

[sleepyp]

Is it possible that higher pressures (greater than 11) and/or increased leaks make it difficult for the machines to accurately determine centrals vs apneas?
Other than brand of machine, that's the only other difference I can find.
The Dreamstation would go up to 18 chasing apneas, whereas the Resmed hasn't gone over 10.3 yet.

[sleepyp]

   
   

Some OSCAR screengrabs for reference.

[sleepyp]

A further update.
The Resmed machine has still not recorded a single apnea event, consistent with my sleep study. The Resmed's maximum pressure has been around 11, so when I travelled with the Dreamstation Go this weekend I used 11 as the max pressure setting, min at 9.5. No apenas in the result (0.59AHI), and a bloody good night's sleep!

I'm quite convinced that at pressures greater than 12 (with 0 leaks) the Dreamstation Go routinely mis-marks CAs as OAs and further ramps the pressure, causing big clusters of OA/CA and very disturbed sleep. Lowering the maximum pressure seems to prevent this from happening. Anyone else experience this?
The Resmed Autoset does not appear to go into runaway in the first place and so I've left it at 20 max.
Might be worth noting for anyone using the Dreamstation Go unit in auto.

[Gideon]

Sleepy, you just described treatment emergent Central apnea, though we usually see it as a result of flex, EPR, or PS increases

[sleepyp]

Sleepy, you just described treatment emergent Central apnea, though we usually see it as a result of flex, EPR, or PS increases

I think so, based on what you and many other knowledgeable people of this forum have said.
Based on this then it's easy for me to set a maximum pressure to prevent the machine doing this.

What I'm surprised by is that a medical device is released on the market that routinely mis-reads events, and compounds these issues by reacting in a way to cause further instances of the same events. I'd have expected that if the device showed this behaviour in testing then it would not receive approval as it is clearly detrimental to the patient. Surely the CA/OA detection testing is done through the full range of pressure of the machine?

[Gideon]

can you post those charts

[sleepyp]

can you post those charts

The charts in post #3 here on this thread are good examples.
The Dreamstation chart is full of OAs, that I now believe are actually CAs.
You can see the Dreamstation responding to these as expected by increasing pressure. On this night it only got to 10.8, but it has gone to 18 in the past and regularly hit 16.
I limited it to max 15 eventually because it was waking me so much.

My sleep study showed no OAs (or CAs) and my Resmed machine has not yet recorded a single OA since I've been using it (about 11 nights now).

[Gideon]

Ok, 2 minute zoom views of the events so we can see what they actually are.

[sleepyp]

Thanks for spending time on this Bonjour.