(05-11-2016, 01:50 PM)sonicboom Wrote: There are different display options for the A10. Additionally, MyAir may be reporting a different percentile for its leak rate - I'm not sure it's reporting the 95% leak rate. It maybe 90 or even 70.
So I spent a teency bit of time looking at my data, comparing with the sleepyhead data.
myAir seems to be doing some sort of long term average or median, or something like that - Only a very horrible sleep will show up as horrible in myAir. The only nights that myAir shows an extreme leak rate were nights where I was over-threshold for much of the night. If you have intermittent, relatively short (less than a few minutes), over-threshold leaks it doesn't seem to show up in myAir. Continuous, low-level leaks do show up. For me this is somewhat useless as I tend to have a good seal, unless I don't, then I really don't - which is what I personally am trying to eliminate.
By contrast, the machine (AutoSet10 set to daily) seems to be doing something like 'worst 95% percentile'. I.E. throw the worst 5% of data away, then report the next. Or maybe the average of meaningful leaks. Or something like that. If I have any meaningful leaks that last any length of time at all, the machine will report it and the number will seem to be artificially high. I've had nights that the machine reported at 10-12 or so, and myAir said 1 or 2. So, it seems to give you a good idea of the severity of the leaks you're having, but it for me doesn't seem to give you an idea of how many of them I had.
I personally like the data from sleepyhead the best. I can *see* what is going on over time, and see if it matches my memory of the night. (I.E. never really got a good seal all night, but it wasn't that high level, VS repeated mask shifts and high level leaks). This is a lot more useful to me than some averaged number which I can only guess what it represents.
Take this for what it's worth - I don't have enough nights experience to be able to completely understand the numbers. I do agree with others that the data is accurate from all of the sources - it just seems that the statistical methods applied to come up with the numbers is extremely different.