SleepyHead & SleepMapper Leak Stats Don't Agree

Please don't jump on me about SleepMapper being junk/joke & how much better SleepyHead is.

I'm new to this & just trying to understand what is happening; what numbers mean, etc so that I can discuss my therapy with my doctor, who seems to be an expert at blowing smoke . . . or, at least, implying (by word & deed) that I should just "strap up every night & come back to see me again next year". Obviously, that is an exaggeration, but I'm trying to learn enough to discuss things, rather than "yes, doctor, whatever you say".

Last night was my 1st night using a chinstrap with my full face mask. I slept very comfortably & soundly. I did not feel leakage. It seems to me that I probably should/would be happy with those numbers (per SleepMapper) every night.

SleepyHead (on left side) shows Large Leak = 35.16%

SleepMapper shows % of Night in Large Leak = 70.2% & Mask Fit = 30%

Also, what does the shaded area at extreme left & right of the Event Flag graph mean?

Robear wrote:Please don't jump on me about SleepMapper being junk/joke & how much better SleepyHead is.

*shrug* facts is facts? but, whatever.

Robear wrote:I'm new to this & just trying to understand what is happening; what numbers mean, etc so that I can discuss my therapy with my doctor, who seems to be an expert at blowing smoke . . . or, at least, implying (by word & deed) that I should just "strap up every night & come back to see me again next year". Obviously, that is an exaggeration, but I'm trying to learn enough to discuss things, rather than "yes, doctor, whatever you say".

perhaps you should say "ok" and just never go back.

Robear wrote:Last night was my 1st night using a chinstrap with my full face mask. I slept very comfortably & soundly. I did not feel leakage. It seems to me that I probably should/would be happy with those numbers (per SleepMapper) every night.
SleepyHead (on left side) shows Large Leak = 35.16%
SleepMapper shows % of Night in Large Leak = 70.2% & Mask Fit = 30%

unfortunately, I have no idea what sleepmapper wants to call 'large leak'.

the machine writes a 'large leak' flag into the data stream that it saves on the card, sleepyhead reads that, and displays it. it looks like if you zoom out, that the LL flag that's shown would be around 35% of the night.

Robear wrote:Also, what does the shaded area at extreme left & right of the Event Flag graph mean?

that you zoomed in, and the charts below aren't showing those parts of the night.
please don't shrink your pics before sending to imgur, your sleephead screenshot is too small to read the numbers. please follow the instructions here on how to upload full size pics: https://sleep.tnet.com/reference/tips/imgur

First the easy question:

Robear wrote:Also, what does the shaded area at extreme left & right of the Event Flag graph mean?

You have selected the part of the flow rate curve to display to be that part of the night that lies between 00:20 (12:20 am) and 6:10. The gray shaded areas to the left and right in the Event Flag table indicate that those parts of the night are NOT currently being shown in the detailed graphs. In Sleepy Head you can zoom on on any part of the night, down to seeing stuff in increments as little as 30-60 seconds if you want to. The grayed out areas on the event table help you locate what part of the night you are zooming in on.

Now for the harder questions:

I'm new to this & just trying to understand what is happening; what numbers mean, etc so that I can discuss my therapy with my doctor, who seems to be an expert at blowing smoke . . . or, at least, implying (by word & deed) that I should just "strap up every night & come back to see me again next year". Obviously, that is an exaggeration, but I'm trying to learn enough to discuss things, rather than "yes, doctor, whatever you say".

It's a shame that your doc is not more open minded about educating his patients. Been there and done that with (multiple) sleep docs until I found one who likes the fact that I look at my data and understand what it means.

You might find the Understanding the data in your sleep study in my signature a good place to start with learning what all the numbers mean.

Last night was my 1st night using a chinstrap with my full face mask. I slept very comfortably & soundly. I did not feel leakage. It seems to me that I probably should/would be happy with those numbers (per SleepMapper) every night.

SleepyHead (on left side) shows Large Leak = 35.16%

SleepMapper shows % of Night in Large Leak = 70.2% & Mask Fit = 30%

I don't know why there's such a huge discrepancy between the Large Leak numbers in SH vs. SleepMapper on this particular night. However, there is a huge and long Large Leak that is flagged in SH: It appears that the Large Leak is about 2.75 hours long (ie 2 hours and 45 minutes long). SH says that you total usage for the night is just under 8 hours, and so the SH "time in Large Leak" should be about (2.75/8)*100% = 34%. So the SH computation for time in Large Leak is correct, given the data shown in the detailed data graphs. The SleepMapper website says that you spent 5.5 hours in Large Leak (i.e. 5 hours and 32 minutes). Since the SleepMapper website does not show you any of the detailed software, we cannot determine why SleepMapper thinks your Large Leaks lasted so long. It would be might interesting to see what Encore Basic or Encore Pro shows for this night in its detailed graphs by the way.

Now if we just focus on the Leak Graph in SH, we get a big clue as to what might be going on. In the following screen shot, I've outlined the official Large Leak in a yellow box. I've also outlined a very suspicious area of the leak graph in a purple box:



Note that the leak rate in the purple box is just as high as the leak rate in the yellow box. And if we add the times for those two boxes together, we get 4.75 hours (4 hours and 45 minutes). If we assume that the leaks just before this part of the detailed data were just as bad for about 15 minutes more, we see that the leak rate is up in that Large Leak territory for right about 5 hours, which matches the SleepMaster number quite nicely.

And that begs the question: Why is that purple box NOT flagged as a large leak in SH? Well, as far as I know, SH uses codes in the detailed data that show up in Encore Pro as "Large Leaks". And that's why I'd love to see what this night looks like in Encore Pro. (Encore Pro is the official PR software designed for sleep doctors and DMEs to use to monitor patients with System One machines that do NOT have a wireless modem.)

And that begs the question: Does SleepMapper flag Large Leaks in a way that is different from the way that Encore (and hence SH) flag Large Leaks?

Also out of pure curiosity: Did you check the machine's LCD for this night? If you did, what did the machine's LCD report for the %Large Leak?

palerider wrote:perhaps you should say "ok" and just never go back.

I'm thinking seriously about doing that, but I have a heck of a lot to learn 1st. I don't want to jump from frying pan into the fire. Won't I need "doctor's orders" for Medicare & BlueCross to pay for supplies?

palerider wrote:unfortunately, I have no idea what sleepmapper wants to call 'large leak'.
the machine writes a 'large leak' flag into the data stream that it saves on the card, sleepyhead reads that, and displays it. it looks like if you zoom out, that the LL flag that's shown would be around 35% of the night.

SleepMapper's data comes from the card also. It seems logical (to me) that it would be handled the same by both software . . . obviously, not. In fact, it looks like SleepMapper reversed Mask Fit & Large Leak.

I'm under the impression that achieving AHI of 0.63 is excellent, assuming there aren't other problems, either during my sleep or during the daytime. Why should I even be concerned about leaks?

Any zoom-in or zoom-out was purely accidental on my part. I don't know how to zoom.

I've seen it mentioned several times that it is possible to zoom sleepyhead "down to a single breath". How do I do that?

palerider wrote: please don't shrink your pics before sending to imgur, your sleephead screenshot is too small to read the numbers. please follow the instructions here on how to upload full size pics: https://sleep.tnet.com/reference/tips/imgur

Actually, I looked for "size" on imgur, but didn't see it. I think what happened was that I had increased the font size on another site & didn't set it back, so it was off the page. Sorry, I'll do better in the future . . . hopefully . . . I've been known to forget.

Are images actually downloaded from imgur into posts or just the links? I'd prefer deleting them from imgur if they are no longer needed.

Robear wrote:

palerider wrote:perhaps you should say "ok" and just never go back.

I'm thinking seriously about doing that, but I have a heck of a lot to learn 1st. I don't want to jump from frying pan into the fire. Won't I need "doctor's orders" for Medicare & BlueCross to pay for supplies?

The script should have indicated a "life-time need" or at least a very extended "renewal period" like 99 months. Once the DME has the script on record, you buy supplies directly from them without the need for your doctor to order things like a new mask or a new hose.

Robear wrote:

palerider wrote:unfortunately, I have no idea what sleepmapper wants to call 'large leak'.
the machine writes a 'large leak' flag into the data stream that it saves on the card, sleepyhead reads that, and displays it. it looks like if you zoom out, that the LL flag that's shown would be around 35% of the night.

SleepMapper's data comes from the card also. It seems logical (to me) that it would be handled the same by both software . . . obviously, not. In fact, it looks like SleepMapper reversed Mask Fit & Large Leak.

I would NOT make that conclusion based on this data. Go back and read what I just wrote.

PR has never been very transparent with what triggers a "Large Leak" in their machines. We know from the backwards engineering that JediMark and others (including palerider) have done that when a System One machine flags a Large Leak, there's a code that is written to the card. What triggers the machine to write a Large Leak flag to the card is NOT known.

In this case, the fact is: The total leak rate in the two hours before the start of the flagged Large Leak is just as large as the total leak rate in the official Large Leak. And that's troubling: It means that we really don't know why those leaks are not official Large Leaks, but the ones of the same size are Large Leaks. This is a problem that originates with Philips Resprionics and NOT SleepyHead.

Now it's just possible that the SleepMapper software looks at the raw total leak rate data rather than the Large Leak flags when it is determining its "Large Leak Time". SleepMapper clearly uses its "Large Leak Time" to then compute the "% of night in Large Leak" figure that is also reported in SleepMapper.

That's why this night is a really interesting piece of data. And why it would be very useful to see what it looks like in Encore Pro or Encore Basic. If SH agrees with Encore, but SleepMapper doesn't agree with Encore, then PR has changed how they've programmed the SleepMapper software to compute "Time in Large Leak", and that must mean there's some criteria that SleepMapper is using other than the Large Leak flags written to the card by the System One machine.

I'm under the impression that achieving AHI of 0.63 is excellent, assuming there aren't other problems, either during my sleep or during the daytime. Why should I even be concerned about leaks?

Because Large Leaks mean that the AHI data may not be accurate. When the leaks get into Large Leak territory and stay there for any period of time, the machine has a tough time accurately tracking the breathing. And if it can't accurately track the breathing, it can't accurately score events. It may miss events that actually happen. Or it could record events that are not really there. Either way, the data becomes unreliable.

And even the 35% Time in Large Leak is a significant amount of time in Large Leak territory. To put it in perspective, on the old Resmed S9's you get a Red Frowny Face on the "Mask Fit" data when the time in Large Leak is greater than 30%.

Any zoom-in or zoom-out was purely accidental on my part. I don't know how to zoom.

I've seen it mentioned several times that it is possible to zoom sleepyhead "down to a single breath". How do I do that?

Click and hold the mouse on the flow rate graph and drag to the end of the part you want to zoom in on. Then release the mouse. SH will zoom to the selected chunk.

Or put the mouse where you want to zoom and and simply click and release. Each click and release will zoom in by a fixed percentage. Click and release multiple times until the graph is zoomed in as far as you want.

Once the graph is zoomed in, you can scroll through the detailed graph by right-click and drags: Hold the right mouse button down while you drag in the direction you want to go.

Are images actually downloaded from imgur into posts or just the links? I'd prefer deleting them from imgur if they are no longer needed.

Just the links are put into the post. When you remove the image from imgur, it no longer shows up in the post.

Robear wrote:

palerider wrote:perhaps you should say "ok" and just never go back.

I'm thinking seriously about doing that, but I have a heck of a lot to learn 1st. I don't want to jump from frying pan into the fire. Won't I need "doctor's orders" for Medicare & BlueCross to pay for supplies?

doesn't have to be THAT doctor, though

Robear wrote:

palerider wrote:unfortunately, I have no idea what sleepmapper wants to call 'large leak'.
the machine writes a 'large leak' flag into the data stream that it saves on the card, sleepyhead reads that, and displays it. it looks like if you zoom out, that the LL flag that's shown would be around 35% of the night.

SleepMapper's data comes from the card also. It seems logical (to me) that it would be handled the same by both software . . . obviously, not. In fact, it looks like SleepMapper reversed Mask Fit & Large Leak.

maybe, and after reading RobySue's analysis, maybe there's a bug in sleepyhead...

would you take your SD card, zip it up, and either upload it to dropbox, or PM me and I'll send you my email address... i'm working with Jedimark right now on the next version of SH.

Robear wrote:I'm under the impression that achieving AHI of 0.63 is excellent, assuming there aren't other problems, either during my sleep or during the daytime. Why should I even be concerned about leaks?

if the leaks are too large, the machine can't tell what kind of apneas you are having so reporting is less accurate. if they're really huge, you lose effective therapy pressure.

Robear wrote:I've seen it mentioned several times that it is possible to zoom sleepyhead "down to a single breath". How do I do that?

as mentioned above in my post, either use the mouse, highlighting a section zooms to it, left click zooms in, right out, or use the arrow keys.

Robear wrote:Actually, I looked for "size" on imgur, but didn't see it. I think what happened was that I had increased the font size on another site & didn't set it back, so it was off the page. Sorry, I'll do better in the future . . . hopefully . . . I've been known to forget.

Are images actually downloaded from imgur into posts or just the links? I'd prefer deleting them from imgur if they are no longer needed.

the instructions are all on that page I linked.

posts are on imgur, and only show up when someone reads an article, no pics are stored on cpaptalk, which is why you need imgur, or other image hosting (imgur is best)

Actually, I'm beginning to question SleepMapper data more & more. They changed the display screen (slightly) in the past couple of days. They now show a zoom box for 14/30/60/90 days, but clicking it does nothing. Maybe it'll have to run x days before each of the zooms will work.

SleepMapper data that I posted above shows Best Mask Fit was Aug 8. My SM Mask Fit was 100% for 21 of the 1st 22 days of 2016 . . . 97% on the other day. Maybe Aug 8 was the 1st 100% night. On 2/6/16 their Mask Fit time & % "did not compute" with their Large Leak time & % . . . % + % = 100%, but time + time did not equal usage time.

If I had not (finally) seen/felt the positive results of CPAP, I would probably just quit. In fact, I surprised myself by sticking it out for almost 6 months with all kinds of problems. Then after a 3 week absence everything suddenly seemed to miraculously "fall into place". It wasn't perfect, but so much better. That's when I started trying to track my numbers, check forums, etc. For some unknown reason, my supplier is 3/4 of the way across the country. I just waited over a week to get a chinstrap. That will definitely change after my dr appt in 2 weeks.

robysue wrote:Note that the leak rate in the purple box is just as high as the leak rate in the yellow box. And if we add the times for those two boxes together, we get 4.75 hours (4 hours and 45 minutes). If we assume that the leaks just before this part of the detailed data were just as bad for about 15 minutes more, we see that the leak rate is up in that Large Leak territory for right about 5 hours, which matches the SleepMaster number quite nicely.

And that begs the question: Why is that purple box NOT flagged as a large leak in SH? Well, as far as I know, SH uses codes in the detailed data that show up in Encore Pro as "Large Leaks". And that's why I'd love to see what this night looks like in Encore Pro. (Encore Pro is the official PR software designed for sleep doctors and DMEs to use to monitor patients with System One machines that do NOT have a wireless modem.)

And that begs the question: Does SleepMapper flag Large Leaks in a way that is different from the way that Encore (and hence SH) flag Large Leaks?

not anymore.... this bug has been caught, and killed. Sleepyhead now reports numbers that are consistent with what sleepmapper and encore pro show... 70% of the night with large leaks.

palerider wrote:

robysue wrote:Note that the leak rate in the purple box is just as high as the leak rate in the yellow box. And if we add the times for those two boxes together, we get 4.75 hours (4 hours and 45 minutes). If we assume that the leaks just before this part of the detailed data were just as bad for about 15 minutes more, we see that the leak rate is up in that Large Leak territory for right about 5 hours, which matches the SleepMaster number quite nicely.

And that begs the question: Why is that purple box NOT flagged as a large leak in SH? Well, as far as I know, SH uses codes in the detailed data that show up in Encore Pro as "Large Leaks". And that's why I'd love to see what this night looks like in Encore Pro. (Encore Pro is the official PR software designed for sleep doctors and DMEs to use to monitor patients with System One machines that do NOT have a wireless modem.)

And that begs the question: Does SleepMapper flag Large Leaks in a way that is different from the way that Encore (and hence SH) flag Large Leaks?

not anymore.... this bug has been caught, and killed. Sleepyhead now reports numbers that are consistent with what sleepmapper and encore pro show... 70% of the night with large leaks.

So I take it that Encore agreed with SleepMapper on this night?

It's good you caught the bug and killed it. I'm just glad that I was able to ask the appropriate question that led you to track that bug down and swat it hard.

robysue wrote:So I take it that Encore agreed with SleepMapper on this night?

It's good you caught the bug and killed it. I'm just glad that I was able to ask the appropriate question that led you to track that bug down and swat it hard.

yes. you are, and were, completely correct. I have to admit that I saw what you did, the previous high leak rate, and wondered about it, but I put that down to my not being as familiar with exactly how, and when respironics flags large leaks. It's to your credit that you analyzed it and pointed out the seeming discrepancy, and extrapolated what probably should have been flagged.

Kudos to Robear for putting up with my pestering to get me a copy of his data, so that I could run it through Encore Pro and mark and I could look at it.. what that, Mark was able to find the bug, and squish it flat

(though he did leave the mess for me to clean up afterwards... bug guts, ewww)

palerider wrote:not anymore.... this bug has been caught, and killed. Sleepyhead now reports numbers that are consistent with what sleepmapper and encore pro show... 70% of the night with large leaks.

Does that mean that I (& others) should download & install a new version of SleepyHead? Should my "production version" of SleepyHead be an older version; NOT the most recent? If I had abandoned my recording of SleepMapper data, like a number of people here have strongly (& IMHO sometimes rudely) recommended to me & others, I would be "fat, dumb & happy" & never would have stumbled upon this bug.

I just clicked on "Help" & then "Check for Updates" in SleepyHead . . . nothing happened ! ! !

Robear wrote:I just clicked on "Help" & then "Check for Updates" in SleepyHead . . . nothing happened ! ! !

it's in development. you can download the source and compile the current dev version if you want the update right now.

These other folks know way more than I but I wanted to mention that the algorithm does not consider the time that you are in Large leak as treated and that time period is not counted in your AHI. Now this has been consistent with my Resmed machine results. Haven't confirmed with PR machines.

Sew tired and can't get logged in wrote:These other folks know way more than I but I wanted to mention that the algorithm does not consider the time that you are in Large leak as treated and that time period is not counted in your AHI. Now this has been consistent with my Resmed machine results. Haven't confirmed with PR machines.

which "algorithm"?

the AHI calculation has nothing to do with large leak. it's events divided by hours.

palerider wrote:

Sew tired and can't get logged in wrote:These other folks know way more than I but I wanted to mention that the algorithm does not consider the time that you are in Large leak as treated and that time period is not counted in your AHI. Now this has been consistent with my Resmed machine results. Haven't confirmed with PR machines.

which "algorithm"?

the AHI calculation has nothing to do with large leak. it's events divided by hours.

On the PR machine I know this happens in Encore:

If a Large Leak is flagged, but breathing is still detected, the time counts as "therapy/usage hours".

If there's an actual "patient disconnect" where the leak is so large that the breathing can no longer be detected, but the machine is on, that does not towards "therapy/usage" hours.

In an Encore printout, the "Breathing Not Detected" shows up as an ugly Black Bar over the Leak Rate graph. A Large Leak shows up as a light Green Bar as I recall in an Encore report. (I really don't want to have to go into Parallels and launch my current version of Encore Pro to confirm that light Green Bar thing right now.)

As I recall, PR uses the therapy/usage hours to compute the AHI. In other words, as long as breathing is detected, that counts as part of the denominator. If breathing is not detected, that time is NOT used when computing the AHI.

I never had an official Large Leak during the three or four months that I used a Resmed S9 way back when. So I have no idea what the Resmed machines do if there's an official Large Leak with breathing detected. If you or JediMark have any Resmed data lying around that has some extended Large Leaks, you might want to see how that data looks in ResScan vs. SleepyHead.