Calculating AHI

I am using SleepyHead to review my data. I am a bit confused on how the software calculates daily and periodic AHI. One specific question is what does it use for ET? Does it use the time the machine is on? Or is it smart enough to subtract out the time the mask falls off, which can be detected by a huge leak reading?

I have noticed some odd behavior in the latest beta version on the way AHI is calculated over multiple sessions. If you select a single session, it gives the AHI number as the number of events in that session divided by _at least_ the total of all session times rather that just that session. I don't know if this is intentional or not. It has the property that if you have two sessions, the total AHI is the some of the two sessions AHI, but it does not give an idea of the rate of events for an individual session. However, there is a strip chart for that...

I don't know if the gaps between sessions are included or not, I have been meaning to check. Sleepy head will give you the total number of events and the session times, so you can do your own calculation.

SleepyHead and Encore both calculate AHI for the entire night in the same way.
Example...You go to bed and turn on the machine and put the mask on at 10 PM but you don't go to sleep until 11 PM...the machine doesn't know if you are awake or asleep but it know if you are attached to it...so that hour counts any any calculations...
So you fall asleep at 11 PM and wake up at 3 AM and turn off the machine and get up to pee or get a drink or whatever and you don't go back to bed until 4 AM...that hour is not part of the calculations for AHI. It's also not part of the "sleep time".
So that would be 1 session that lasted from 10 PM to 3 AM... 5 hours
Second session would be if you put the mask back on at 4 AM and don't remove it until 7 AM... so 3 hours
Total usage time 10 PM until 7 AM minus the time from 3 to 4 AM...so 8 hours total.
It is 8 hours that is used to calculate the overall AHI for the entire night.

If a leak is so big that the machine can't tell you are attached to it and it remains on...it will report "breathing not detected". Big blank spot on the graphs...I don't know if that time is part of the AHI calculations or not. Actually never thought much about it because if someone has that big of a leak for a prolonged of a period of time...they got more things to worry about than minor variances in the AHI that would happen with that time being included or omitted.
If it's a short lived big leak where breathing isn't sensed..whether it is included in the calculations or not isn't going to make much difference in the overall AHI one way or the other.

I have never had a leak so big that the machine couldn't tell I was attached to it and breathing...if I did then I could manually calculate the AHI and figure out if that time is part of the calculations or not. I don't think it is but I haven't had a chance to actually put a pencil to it to prove it.

But if the machine is turned off when mask is removed or if a person uses the auto off so that it turns itself off...with the PR S1 that is 60 seconds and that 60 seconds isn't going to impact the AHI at all. It won't even affect the overall leak numbers beyond given a higher maximum hit.
ResMed machines will auto off even sooner and again no real difference in leak numbers beyond the max hit because the time is so brief.

It sounds like if the mask is not on your face -- fell off or your forgot to turn off the machine when you got up -- that time is subtracted from total time of usage (the AHI denominator).

So, if one went to bed at midnight. Mask fell off at one. Put mask back at two. Got up at seven. But left the machine run until eight. You would have eight hours of runtime, but six hours of sleep time used to calculate AHI? Correct?

What events go into the numerator?

pgflrob wrote:So, if one went to bed at midnight. Mask fell off at one. Put mask back at two. Got up at seven. But left the machine run until eight. You would have eight hours of runtime, but six hours of sleep time used to calculate AHI? Correct?

Midnight to 1 AM when mask fell off....so one hour of usage that will be part of any AHI calculations...if the mask is off and machine still blows..that is where the "breathing not detected" comes into the picture.
the time between 1 AM and 2 AM where the machine runs but a human is attached...just increases the blower hours and isn't counted towards usage hours and thus not part of any calculations.
Mask back on at 2 AM until 7 AM..5 hours of use.

So yes...5 plus 1 is 6 hours of usage or sleep time (may or may not be asleep) which is used to calculate the hourly indexes.

pgflrob wrote:What events go into the numerator?

Numerator for what? The AHI?
AHI is composed of 3 indexes (hourly averages)....the Clear Airway (Central) Index...the Obstructive Apnea Index and the Hyponea index.
The other flagged events on the events graphs...are not part of the AHI.
They happen and can be important but are not included in the AHI.

To keep it simple.
10 Clear airway events over 6 hours
10 Obstructive apnea events over 6 hours
10 Hyponeas over 6 hours

That's 30 events over 6 hours and the overall AHI is 5...30 events in total divided by 6 hours
Each individual index would be 10 divided by 6 or 1.6666xxxx.
The overall AHI is nothing more than the total of the 3 event categories.