SleepyHead Qt v0.8 alpha

[jedimark]

Hmm..

The main reason I find it useful, is I have a habit of masking up and coding for an hour or two on my couch before going to sleep.
With this feature, I can leave off the time I know I'm awake, and get a more accurate measure.

Jimbo also made the suggestion to me of hiding the "AHI" figure when past a certain zoom level.. Could always just show AHI=?? when too close to be useful?

On another tangent: If REM cycles are an average of 90 minutes, I wonder why AHI is calculated per hour anyway?

One day I'll get around to adding highlights/bookmarks.. It would likely start as a list view in the left pane to select or delete individual bookmarks, and a menu option/button to add the currently selected area as a bookmark, along with a small attached note or category marker.

Maybe an extra option/category to flag an area as "awake" and have the area excluded from all calculations?

Bookmarked areas would show over the flow waveform as colour coded rectangles, like PB/CSR does.

[jedimark]

I was just having a further look at the ResMed ASV data ignorant1 sent me.

It's importing fine, so I'm fairly sure those last VPAP bug fixes I made were post 0.8.6

There is one new field for this ASV, and I'm not sure what it might be:
08868: Unobserved ResMed PLD Signal "TgMV"
(SleepyHead skips over unknown S9 signals, and reports them like the line above in the Help->Debug window)

Can anyone give me a clue what this "TgMV" might be? It looks like a cleaner, filtered version of Minute Ventilation.
Does anyone know if ResScan gives a proper name for it?

I just added it to the S9 import module, and created a new graph called TgMV for now..

[ignorant1]

I’ll take a stab at this:

Seems likely that TgMV is the data used to calculate “Target Minute Ventilation”.

Target Minute Ventilation is a variable that the ASV algorithm adjusts, and it is shown only in the “Summary” section of ResScan as a numerical value (not as a tracing anywhere else). It is displayed in a similar manner as “Minute Ventilation” with a Median, 95th percentile, and Maximum value reported.

Or does this show up on other VPAP units too? (I don’t have any experience with BiPAP – only straight APAP/CPAP and now ASV - hence my moniker…)

[archangle]

Hmm..

The main reason I find it useful, is I have a habit of masking up and coding for an hour or two on my couch before going to sleep.
With this feature, I can leave off the time I know I'm awake, and get a more accurate measure.

I wonder if it would make sense to simply list a count of events in the window, along with the duration of the window. Maybe add events/hour as a third column. Perhaps as a new tab equivalent to Details/Events/Notes.

[Breathe Jimbo]

Maybe an extra option/category to flag an area as "awake" and have the area excluded from all calculations?

Great idea! Strange, unusual things happen as we fall asleep and wake up.

[jedimark]

A new Windows test build of SleepyHead (0.8.7) has just been uploaded. Get it while it's hot!

Still pre-Beta.. Hopefully it's getting a little less breaky..
Please keep them bug reports and feature requests coming at sourceforge.

For the full version, with DLL's, simply download and unzip.
http://sourceforge.net/projects/sleepyh ... p/download

Alternatively, if you've already got the last build, just grab this binary, and unzip the SleepyHead.exe binary over the top of the old one.
http://sourceforge.net/projects/sleepyh ... p/download

Jimbo said he will build an updated mac version soon, so keep an eye on the mac thread for his post..
(Thanks again Jimbo for all your hard work on the Mac testing and building front.)

If there's any Linux (or BSD) users who are finding it difficult getting sleepyhead to build or run, please let me know.

What's new?
Most of it is now explained in the release notes on first start, but here is a quick rundown

Can now Shift+Click on an overview graph to jump to that day in daily view.
Better S9 ASV/VPAP support, hopefully less crashy now.
Basic CSV export capabilities.
Some daily view calender navigation improvements.
"Live" event count, duration, "ahi" calculations shown in Flow Rate waveforms.
Some graphing optimisations and performance improvements.

Catches & Known issues.
Starting day of week is hardwired to Sunday in this version.. The Locale fix to this is in the code already, but it requires Qt 4.8 to work, and that's not officially released yet. (This is built with 4.7.4) Apologies in advanced to people who get messed up by this.. It will be built with 4.8 next time around, fixing this.

Daily printing still isn't done.
Oximetry is still flakey..
Both those two are waiting for a fog free day.

Well have fun, I hope this is a little better than the last one..

/Mark

[theSleeper]

Hi Mark,

Just moved from 0.8.5 to 0.8.7. Looks good !. Excellent job.
Crashes from daily view to overview (after download) are gone.
More features (I like the setting for mask leakage).
Export function is great. I still keep track of each night in an excel sheet with nice graphs but I copy the data after reading them using SH. It may be time to change that to allow automation.
Nice to see that you can pick up a graph on the left hand text and move it up and down (change display order)

Will try with oximeter data later. Still need to figure out how the life data works there (compared to getting recorded data).

Again, thanks !.

Sleeper.

[ignorant1]

I just downloaded the latest version... and SUCCESS! The software now imports my Resmed S9 ASV data without any problems.

I've only played around with Sleepyhead for a few minutes, but WOW! This is GREAT! It loads the graphs almost instantaneously, which is such a contrast to the ResScan bloatware. I'm really looking forward to playing around with it and discovering some of the other features & functionalities when I get back later tonight.

Kudo's to Mark for all his hard work - and for allowing us to have such a great alternative to the manufacturer's software!

[Pugsy]

It loads the graphs almost instantaneously, which is such a contrast to the ResScan bloatware.

Yep hugely faster. For those wondering about how it compares with Encore Pro...
I can open Encore Pro..do the log in and password thing...and then Pro starts to load...little circle spins...
I then can go to SleepyHead..open..download and see the report...and then go back to Pro and it will just be finishing up loading the software to the Overview screen. I kid you not. This doesn't even count the time need to generate the reports which is excessive. The only thing sort of quick in Pro is the downloading of the data from the SD card.
Pro 2.4 is slow...Pro 2.5 is even slower than 2.4... Even Encore Viewer 2.0 is slow compared to SleepyHead.

[robysue]

Glad you're doing better. Take care of number 1 first.

Oooh, the more I think about instantaneous AHI, the uglier it gets. If you sit in one spot and zoom in, the AHI number will go up. which I think will be confusing. Not calling it "AHI" is definitely a good idea. As a silly example, if you look at a 1 minute window, the only possible values would be 0, 60, 120, 180, etc.

Since the apneas and hypopneas are discrete events with measurable time between them (even if it's short in a nasty cluster), mathematically the idea of an "instantaneous AHI" makes no sense at all.

I wonder what number the manufacturers report if, for instance, you run the machine for 2 minutes and have one apnea? Do they report it as AHI=30, or just don't report?

I have no idea what the manufacturers' do about this. But we know the consensus of the sleep medicine industry on measuring specialized AHI's involving short periods of sleep time in full NPSGs: You count the total number of events and divide by the time elapsed measured in hours regardless of how small that time might be. For example, if you have a scant 3 minutes of REM on a diagnostic PSG and you have ONE apnea or hypopnea during those 3 minutes, you wind up with a test result that says REM AHI = 20. Now, is that extrapolation based on a single three minute REM cycle really typical of what goes on in your untreated REM cycles night after night? God only knows, but the presumption is that the extrapolation means something in terms of OSA, where in reality there is a possibility that the lack of REM for some reason other than OSA and that that particular three minutes of REM were atypical. Of course, for most of us, there's enough NREM data with bad AHI numbers that we know we have a problem with OSA regardless of what really happens if/when we get into REM ...

And computing AHI numbers based on significantly less than an hour's worth of data DOES create problems with understanding the data coming out of our NPSGs: Judging on the sleep report data routinely posted both here and on apneasupport.org's Sleep Studies subforum, it's pretty common for folks with OSA to get truly tiny amounts of REM on that diagnostic PSG. As in 5 minutes or less of REM. And yes, the REM AHI is computed by dividing the number of events by the REM time measured in hours. And that's how it's possible to come up with truly frightening AHIs during REM: If you have a scant 3 minutes of REM and you have 2 events during those 3 minutes, you wind up with a test result that says REM AHI = 40. And folks posting about their data often have questions asking "How is it possible to have 40 events in 3 minutes of REM?"

Think about what it would look like if you graphed this "short term AHI value". Since the AHI index is based on a low number of discrete events spread over a fairly long time period, the idea of short term AHI numbers gets complicated. You almost need something like a shaped time window like an exponential decay function.

And x-axis measures what? The time interval? In that case, the graph would look more like y=1/x with a vertical asymptote at x=0 if an event occurs at the start of the time interval. (One event divided by 0 hours equals positive infinity.)

If you do just a 1 hour square sliding window, it will jump up by 1 every time there's an event, and even more confusingly, drop 1 unit one hour after every event. If you have a single event or a small cluster of events, it will jump up and stay up for 1 hour after the events have stopped.

In ResScan there is a graph of the AHI. And it does indeed jump up by 1 every time there is an event. But it's not a sliding one hour window: At the beginning of every hour, the AHI is reset to 0. So you get a graph that marches upwards counting the events that occurred since **:00 and then the graph suddenly drops back to 0 precisely one hour later. I never saw much point in that graph myself.

I'm not necessarily suggesting implementing a complicated function in SH to calculate that. I'm just thinking that the math and philosophy involved is complicated. I'm not sure there's a good answer.

I'm just joining in the mathematical discussion about what the idea of AHI when computed over short periods of time actually means.

[jedimark]

I really intended the "live" AHI thing just for an interim way of ignoring time spent wide awake doing stuff with the mask on, which dilutes real AHI calculations.

This "AHI" value zoomed in real close doesn't mean much at all. I could simply make this "AHI" value disappear when zoom level is less than 1 hour, and just show the event count & duration.

I do see one potentially useful case for it: When looking at a half-hour or so sized block of PB, with clear-airway events happening in between each cycle. It may help users simplify other calculations.

It would probably make a lot more sense if sleep stage was overlaid or at least comparable.

I think the ability to bookmark sections, and be able to flag them to "Ignore", will sort out the dilution issue nicely.

Anyway, you guys now have the chance to play with it.. maybe one of you will find an even better use for it.. If not, it doesn't have to stay..

[theSleeper]

Keep in mind that AHI is just an indication. If I see that for 3 hours in a night I didn't have any event or I got a number of longer period without events, it means I had a good night as I didn't wake up often and probably went through most sleep stages. If there was an AHI of 8 that night, it was outside of that "quality" time. An AHI of 5 where the events are nicely spread, one each 12 min, would probably result in you never reaching all sleep stages.
I think it makes sense to be able to check the AHI easily between points but only longer periods make sense as Mark indicated. I like the idea of at least an hour and otherwise indicating # events for the selected period.
I'm just getting ideas of what I can do with the exported data from SH. In a spreadsheet you could easily track the longest periods per night without events. This could be an additional parameter to quantify the quality of your sleep.

[jedimark]

I just added two new features on SleepyHeads project website on sourceforge.

One is a Wiki, which can be accessed as the main home page for now (using either http://sleepyhead.sf.net or http://sleepyhead.sourceforge.net)
And the other is an IdeaTorrent, which is a feature request system, but with voting goodness.

There is no user guide yet, but I did write up some software build instructions, and a couple of other little bits..

[rosacer]

I just opened the new version, very good I like a lot the Review page

Thanks jedymark ! I hope your health is going better.

Rosie

[tetragon]

What was the build environment for the 0.8.7 OS X build? I haven't had a compile succeed since fcbb57ae84ec387a950977ba72b763ca0d7c5936. I'm running on an older 10.6, with an older Xcode, and Qt 4.7.4.