AHI

[Joerg Hickman]

My AHI on the Resmed Apap (Airsense 10) sits between 1 and 3. Whilst I understand what the AHI is, it isn’t entirely clear how the machine arrives at this number. For example, if the AHI is 2 does this mean I had two apnoea/hypopnoea events per hour that it TREATED or does it mean the machine treated MORE than two events per hour (which aren’t reported) but the two events were the ones it didn’t manage to treat successfully? If the former is the case then that would imply that my AHI of 2 is in the normal range and that wouldn’t warrant any treatment.
My guess is that there may be a predictive algorithm involved that anticipates an event and intervenes before the apnoea/hypopnoea criteria are triggered.
Many thanks to anyone who can help clarify this apparent mystery.

[Jeff8356]

This wiki page may help explain AHI better:
http://www.apneaboard.com/wiki/index.php...ndex_(AHI)

Your doctor will likely say you are properly treated if your AHI<=5 but that may not be the case.  Many users need to get their AHI lower for treatment and better sleep.  Everyone is different and posting a chart from OSCAR will tell us a lot.
D/l OSCAR from the black ribbon at the top of this page and install it.  Put an SD card (up to 32GB) into the machine and use it overnight.  Import the data in OSCAR then use F12 to take a screenshot (fn + F12 on macOS) then post it here.

The MyAir app is more of an encouragement thing.

[pholynyk]

Strictly speaking, the machine doesn't treat apneas...

If the pressure is high enough to keep the airway open, then the apneas are prevented

If the pressure is not high enough and the airway closes, then the machine will detect that apnea and then - after you are breathing again - increase the pressure to prevent the next apnea

Of course there are limits as to how fast the pressure is increased, and how high, but that's the general idea - as far as obstructive apneas are concerned

Central or clear airway apneas are a whole different matter. Increasing the pressure won't open an open airway. We need a different type of machine that can create airflow as needed - an Adaptive Servo Ventilator. That's a different pile of money.

[Joerg Hickman]

Thank you all for your responses. Unfortunately they don’t answer my question. I’m pretty familiar with CPAP, Apap and asv devices in terms of what they do and don’t do. I also use OSCAR but that doesn’t help in answering my question. What I want to know is how the AHI they spit out in the morning is derived. I know they all have their proprietary algorithms and I may need to contact the manufacturer to get an answer.
Thanks again.

[Crimson Nape]

To give you a "heads-up", don't expect any answer from the manufacturer. You are not their customer, your DME is.

Your question of the AHI of 2 - This means that these events could not be prevented by the CPAP. There are certain instances, caused many times by body position, that requires a pressure to overcome the event that exceeds the capability of the CPAP. The number "2" would indicate the ones that got away. CPAP is for prevention. Unless you closely analyze your data, you won't know how many it prevented. Think of it as a seatbelt for the airway. A seatbelt doesn't prevent you from getting hurt, it just reduces the amount of injury.

Your guess about the predictive algorithm, I would say, is close. The Philips code seems to be an "after-the-fact" type of response and is slower to respond than the Resmed. A Resmed uses the flow rate and flow limits to predict an impending event, and will increase pressure prior to an event taking place, most of the time.

I probably haven't answered your question to your satisfaction, but I had to try.
- Red

[clownbell]

The calculation is as follows: count the number of obstructive apneas + hypopnea + centrals, add them up and then divide by the number of hours slept. I believe the number of hours is expressed in decimal format; so 7 1/2 hours would be 7.5, 7 hours 20 minutes would be 7.33, and so on.

[sptrout]

"clownbell" showed the standard, and correct, AHI calculation. One of the big things that throws machine calculated AHI in question is the "hours and minutes" asleep each night. The machine has no idea if the user is asleep or wide awake staring at the ceiling. It would need to measure brain waves to correctly determine actual sleep time. So, instead of actual sleep time, the time used for the machine's AHI calculation is the total time that the machine is on from startup to the final shutdown minus any shutoff times in the middle of the night for bathroom trips or other reasons. Obviously, this error in "sleep time" can really throw-off your machine's AHI calculation and always for the better (lower AHI) because it will always report more sleep time than actual. The only exception is if you have no SA events then the sleep time is not needed; zero is zero. Therefore, to really dig deeper into your nightly AHI you need to review all the other data provided, and displayed, in OSCAR. In my case, I usually sleep one to three hours less than what the machine uses so my actual AHI is always worse than shown unless the displayed AHI is zero.

[clownbell]

Of course sptrout is right. I was taking lazy shortcut. The correct parameter is how many hours the machine is on. It can't tell whether you are awake vs asleep.

[Gideon]

Not quite true, events are not logged during ramps, I think that ramp time is not included, can someone verify that.

[Crimson Nape]

I finally found someone's chart that had the ramp turned on. [Link to Chart Here] .
The ramp time is included, but no events are recorded.