Inspiration expiration ratio

[palerider]

Pugsy please look at: http://imgur.com/eENlV6G Compare the yellowed area with the parenthetical value. Exactly six seconds. I see events like this every night, more frequently 8 & 9 seconds, many with substantial flow in them.

you need to zoom out, and look, the fact that you've shown the 6 seconds that SH shows as an event duration doesn't mean that it's an event. the highlighting is a *BUG* in SH.

look up the definition of an obstructive apnea, it doesn't count, until it's at least 10 seconds long.

[Jay Aitchsee]

Well, nice to see you, PR

Ron, I don't know about your machine. But check where it marks the event. Not all machines mark them at the beginning of the next inhale. Some mark them in the middle. If yours is one, it would affect your measure.

And looking at your results, except for that block in the middle of non use. I think you've made progress with the FFM. Most of your events are hypopneas, now, which would suggest you need to raise EPAP (and IPAP using straight CPAP).

[rkl122]

You are assuming

It's not me, it's SH.

that is where the event starts and stops...but to my eye...I can't tell where it stops and stops.
Believe what you want to believe...I am not going to argue to the point with you but the definition of apnea events all include the 10 second thing.

That's why I called the 10 second rule part of the central dogma earlier in this thread.

You can't prove to me that where you think it starts and stops is where the machine thinks it starts and stops.

It's not me, it's SH! PaleRider has an interesting observation. I wonder if this is a bug in this beta version of SH.

Sorry...but not worth arguing over so I won't go there. The machine will not flag a 6 second air flow reduction as any sort of apnea event.

Unless I'm misunderstanding, you're suggesting that SH is misrepresenting event durations. I'd think this should be great cause for concern, because it can inflate AHI, which is the only number that many docs care about. EDIT - OK, retracting. It's not SH that reports to the doc, it's the machine itself. Doh!

Incidentally, I've just looked at the same time snip in the EncoreBasic waveform. As discussed, no way to measure accurate duration there, but the event itself does appear, and the neighboring flow curve has just the same contour as it does in SH.

Thanks for the image instructions. Very helpful. Perhaps part of my problem is I never created an actual Imgur account. Was just using the service without logging in.

-Ron

EDIT - PR, our posts crossed. If so, that's some bug. Ok, understood, y'all are saying that the AHI computed from the SD card is accurate. JediMark, I hope the source is with you.

[palerider]

Well, nice to see you, PR

Ron, I don't know about your machine. But check where it marks the event. Not all machines mark them at the beginning of the next inhale. Some mark them in the middle. If yours is one, it would affect your measure.

And looking at your results, except for that block in the middle of non use. I think you've made progress with the FFM. Most of your events are hypopneas, now, which would suggest you need to raise EPAP (and IPAP using straight CPAP).

resmeds are good about marking the end of an event. respironics are more vague, even looking at the encore readouts (in as much detail as you can, which isn't that much) they tend to poke the flag near the middle of an event.

different versions of SH interpreted, and guessed at that data differently. before the 1.0 beta, there hadn't been much effort to compare the output of encore with SH in detail, so a lot of things slipped through, like LL flags being half the actual amount, and events being marked at somewhat seemingly random places during the event.

[Pugsy]

I thought you did the yellow highlight thing....
None of my SH versions ever gave me any highlights like that when I zoomed in.
Just now checked some 9 OA events from last night...not a one has anything highlighted...and interestingly enough all showed single digits in parentheses...and yet when I went and zoomed in (no highlights) best I could count was I barely made 10 seconds.

I am currently using the Windows SleepyHead version 1.0.0 Beta 2.

I thought you created the yellow highlighted section just to accentuate where you were counting from.

[Jay Aitchsee]

Unless I'm misunderstanding, you're suggesting that SH is misrepresenting event durations. I'd think this should be great cause for concern, because it can inflate AHI, which is the only number that many docs care about. EDIT - OK, retracting. It's not SH that reports to the doc, it's the machine itself. Doh!

Right, and it's why I've Implied a couple times in this thread that we need to be thinking about the specific machines, their limitations, their intended use, and the software that was written explicitly for them. That's what the health professional is going to use. SleepyHead is great and I'd rather use it than ResScan, but SleepyHead is a generic program written to interpret the data output of several different brands of machines which are not standardized and slight differences in data presentation sometimes appear. That's why whenever a question about a particular anomaly comes up, one of the first questions asked is, "did you check it against the manufacture's software?" ResScan, Encore, or whatever.

[Jay Aitchsee]

I thought you did the yellow highlight thing....

I'm using 9.8, windows 10, with an S9 and today I have the highlighted time box, with and without the cursor, but I swear, I don't think it's there all the time.

[palerider]

I thought you did the yellow highlight thing....
None of my SH versions ever gave me any highlights like that when I zoomed in.
Just now checked some 9 OA events from last night...not a one has anything highlighted...and interestingly enough all showed single digits in parentheses...and yet when I went and zoomed in (no highlights) best I could count was I barely made 10 seconds.

I am currently using the Windows SleepyHead version 1.0.0 Beta 2.

I thought you created the yellow highlighted section just to accentuate where you were counting from.

if you move the cursor over the event, on the flow trace, it should highlight the period it (sh) thinks the event lasted.

like so:

[rkl122]

I thought you created the yellow highlighted section just to accentuate where you were counting from.

I did. The Windows snipping tool (free) has a highlighter.

-Ron

[palerider]

I thought you did the yellow highlight thing....

I'm using 9.8, windows 10, with an S9 and today I have the highlighted time box, with and without the cursor, but I swear, I don't think it's there all the time.

it's a "tooltip" (in interface parlance) something that shows up for a time when you hover over something. it shows for a few seconds, then goes away.

[Pugsy]

I'm using 9.8, windows 10, with an S9 and today I have the highlighted time box, with and without the cursor, but I swear, I don't think it's there all the time

I have never had the flow rate graph be highlighted with any color no matter the zooming in unless it was the green PB color from Respironics flagging of Periodic Breathing.
And I have zoomed in on a lot of individual events over the years both from using the Events tab timed choices and my own random zooming in.
And I have used every SleepyHead version Mark has ever come up with...never ever saw anything but the green PB.
Oh well...shrug...nothing I can do about it and it still doesn't change the inability to see exactly when the flow rate becomes an apnea event and when it returns to normal when zoomed in close enough to count the seconds.

"did you check it against the manufacture's software?" ResScan, Encore, or whatever.

Same here but in terms of the actual AHI...I have never had ResScan or Encore tell me anything different in terms of the AHI or any category breakdown when all data files are available.
SleepyHead AHI and the breakdown....have always matched with the brand name software. So SleepyHead isn't deciding on the flagging...SleepyHead is just reporting what the machine flagged.
But when in doubt...always verify with brand name software if at all possible.

[palerider]

Unless I'm misunderstanding, you're suggesting that SH is misrepresenting event durations. I'd think this should be great cause for concern, because it can inflate AHI, which is the only number that many docs care about. EDIT - OK, retracting. It's not SH that reports to the doc, it's the machine itself. Doh!

respironics has *never* been precise in event duration. iirc, encore (the official software) does not report event duration. so the event duration is something made up in sleepyhead..

secondly, the event duration plays no part in the AHI. simply the count of events, and those events are flagged in the machine and written into the data stream on the card. sleepyhead just reports what's there, as far as how many events the machine tags.

[palerider]

SleepyHead is great and I'd rather use it than ResScan, but SleepyHead is a generic program written to interpret the data output of several different brands of machines which are not standardized and slight differences in data presentation sometimes appear. That's why whenever a question about a particular anomaly comes up, one of the first questions asked is, "did you check it against the manufacture's software?" ResScan, Encore, or whatever.

when I was working on collecting data to test 1.0 beta, I did a LOT of meticulous comparison between SH output and both encore and resscan, so for many things, 1.0 is much more 'true' to the official sw than previous versions. (things like looking at waveform data for the start and end of LL flags, PB events, exact placement of the event flag...)

one thing that SH is different on is some of the left hand statistics, because the official software tends to throw out outliers in the data, maybe skipping the first couple seconds as the machine comes up to speed, or only looking at the 99th percentile, and SH doesn't try to get those averages and mins and maxes to match *exactly*. ... and sometimes, watkins says his math is better than the manufacturers... I never saw the raw data, so... i dunno.

[Pugsy]

I did. The Windows snipping tool (free) has a highlighter.

Gotcha. That highlighter I am familiar with.

The problem I have with the whole thing is that I can't definitely say that the flow rate change during the 6 second period is for sure when the machine starts its thinking...this is where the OA begins and ends.
I can't tell for sure where the beginning and ending is.

Mark elected to use some "markers" he found in the data as a close approximation for event duration but they aren't duration. The machine simply doesn't give us event duration...and the fact that the markers add up to something close to what it appears visually to be a flow reduction earning a flag...it's just a coincidence.
Mark has always told us that he never knew what the "markers" were for but he did tell us that it was close to event duration...and it is for the most part.

If your machine flagged a flow reduction...either OA, Hyponea or ClearAirway...it lasted at least 10 seconds.. Plain and simple. The fact that SH is reporting the markers (we never figured out what they really meant) as 6 and it appears that visually it might be 6 seconds...doesn't mean it was 6 seconds.
It has to last 10 seconds to get a flag. There is no away around that little fact.
Did you verify with Encore that it also had a flag at the exact same time of the one you zoomed in on?

[palerider]

The fact that SH is reporting the markers (we never figured out what they really meant) as 6 and it appears that visually it might be 6 seconds...doesn't mean it was 6 seconds.
It has to last 10 seconds to get a flag. There is no away around that little fact.
Did you verify with Encore that it also had a flag at the exact same time of the one you zoomed in on?

looking at what's highlighted in that pic, you see that there's pretty much flat flow before, and after that point, with what looks like a partial breath *in the middle*.

be nice if rick would repost that spot, zoomed out more to where there's actual breathing on either side of the event. then there might be a chance of seeing what actually was happening, and how long it lasted.