Hello Guest, Welcome to Apnea Board !
As a guest, you are limited to certain areas of the board and there are some features you can't use.
To post a message, you must create a free account using a valid email address.

or Create an Account

New Posts   Today's Posts

How does SleepyHead determine the AHI?
How does SleepyHead determine the AHI that it displays? I checked the Beginner's Guide to Sleepy Head but didn't see an answer; I also looked through the tread titles in the Software Support Forum and didn't see anything that suggested the answer was already posted.

Four possibilities that occur to me are:

1. SH directly copies and reports the AHI number that is calculated and displayed by the particular PAP machine one is using;

2. SH calculates the AHI number itself but, to do so, it uses the same AHI-calculation algorithm that is being used by the particular PAP machine one is using;

3. SH uses its own algorithm to calculate AHI but it does so based on the numbers and types of events (OA, CA, H, RERA, etc.) that the particular PAP machine has flagged;

4. SH uses its own algorithm to calculate AHI and to do so it independently assesses and flags what it defines as an OA, CA, H, RERA, etc. based on its own independent assessment of flow curves, etc.

One reason I am asking this question is that I am currently trying to compare the treatment efficacy of two different PAP machines that I have, and to do so with reference to SleepyHead I need to know in general terms what SH is doing. For instance, I've been told that I can't directly compare the AHIs displayed on my two PAP machines because they use different algorithms. Does this same lack of AHI comparability across PAP machines (brands) "infect" SH's calculation of AHI, making it impossible to directly compare the AHIs as displayed by SH? I know, of course, that ultimately the raw data depends on what the particular machine stores and thus uploads to SH, but I'm interested here in what SH does with that data which leads ultimately to its own reporting of the AHI number.

Thank you!
Post Reply Post Reply
I think the answer lies in the definition of AHI. It is the sum of CA+OA+H events divided by the hours of machine use.
Post Reply Post Reply
The machine itself reports the events and therein may lie the problem. I think that the definition of apneas are tight enough that the reporting of apneas would be very close between machines. There are different algorithms for distinguishing CAs from OAs that may give rise to some differences but since AHI is concerned with the sum of the two that should not be a problem.

The problem is probably with hypopneas. The definition leaves a fair amount of leeway and I can not compare AHIs from my Resmed S8 machine to my S9. My AHI dropped immediately when I started using the S9. Resmed altered the requirements for scoring a hypopnea on the S9 to make it more realistic (realistic is my opinion) so the drop in AHI was probably just a matter of changed scoring requirements for hypopneas.

Someone else on the board probably knows how Sleepyhead comes up with its AHI. I am not sure.

Best Regards,

Post Reply Post Reply

Payton, Sleepyhead relies on the machine event flags and data. It makes no interpretation of events, but does sum and divide by session time.
Post Reply Post Reply

I am not trying to be insulting or anything along those lines. That is what I suspected. How were you able to confirm this? I am not as good at web researching as a lot of you and new to internet message boards although I ran one of my own back in the dial-up days but that was a long time ago.

Best Regards,

Post Reply Post Reply
Sleeprider, I too would hope to know where that information comes from. Likewise, I push this question without intending any disrespect. Is what you say formally described somewhere where I can see the details myself?--and if not, how do you know? Again, no disrespect or suspicion intended; it's just that this is an important issue for me so I want to make sure "with my own eyes" about what I'm dealing with.

SleepyHead is wonderful but if what you say is accurate, it means that the program is not quite as useful as I had hoped, in that it does not let one make valid comparisons across platforms (types of machines), at least with respect to either the AHI number as a whole or with respect to the numbers and locations of specific events that it flags. It means, if I understand correctly (please let me know if I'm wrong), that the only way to compare efficacy across platforms is to try to analyze all the wave forms (or other factors) myself. I can't even say "analyze the *events* myself" because what would be flagged as events would be different on the two machines. I'd have to look through the entire set of detailed data and try to quantify every possible event in both machines and then compare them--a very hard task and one for which I probably wouldn't be able to accomplish without devoting a huge amount of time.

Post Reply Post Reply

I don't know, but infer. I have had direct discussions with Mark Watson, and am peripherally aware of the process he uses to program Sleepyhead to interpret machine code to produce events, flow and other parameters reported by SH. The machines produce the code, the software interprets it to the best understanding of the programmer. The same machine data is used by various manufacturer software to report. The code comes from the machine firmware as recorded on the SD card. If the machine does not provide the event flag, it is not created by SH or any other software. There are pieces of machine code that no one but the manufacturer fully understands, but what you see in SH is a decoding of machine data.

All software is reporting software, however, summary data, average, and statistical information that is reported and graphed, may come from calculations in the software, rather than the machine.
Post Reply Post Reply

Forum Jump:

New Posts   Today's Posts

About Apnea Board

Apnea Board is an educational web site designed to empower Sleep Apnea patients.

For any more information, please use our contact form.