What's Jedimark up to?

[Pugsy]

What does the DST bug look like? How does it manifest?

Double sleep sessions...double hours of sleep (sort of)...events totals doubled (sort of) but individual list of events match events....my EPAP pressure is whacko....average leak says 0.0 when I know it isn't because of too many larger leaks.

The other night I had everything completely normal and all item reported as they should. That was the only night. This is in the Beta version only. I also have 0.8.8 version available and it is not affected.
I am using Windows version of SH. Don't know how or if the Mac version is affected.

[Zzzzzzzzzzz...]

What does the DST bug look like? How does it manifest?

For me, I get duplicated sessions out of sync by an hour, with sleep time doubled.

However, switching the "incorrect" session off restores normality.

Me too. And NOW I know why...

Z

[NateS]

What does the DST bug look like? How does it manifest?

For me, I get duplicated sessions out of sync by an hour, with sleep time doubled.

However, switching the "incorrect" session off restores normality.

james1,

What does this mean? How do you switch what off to restore normality.

Can you please elaborate?

Thanks, Nate

[NateS]

Mark,

Between the two different bugs in SH/Win and SH/Mac, I am wondering what to do.

Should I/we keep importing data into SleepyHead with the idea that, when a fix is found, it will go back and retroactively fix the reporting errors?

Or would it be better to just save a daily copy of the data off the SD card somewhere on our computers, and stop all importing into SleepyHead until such time as you are able to fix it, and then only import after the fix? And just import into ResScan for the time being to know how we're doing.

Your program - SleepyHead - is clearly the better program, when it comes to features like Best and Worst AHIs and Best and Worst Settings, the green and red lines in Prescription settings history, etc. so we are of course willing to wait for you, and again wish you the best in dealing with real life problems first!

Best wishes,

Nate

[jedimark]

STR.edf "summary only" sessions are likely causing the double-ups. This has only really been functioning since the beta.

I vote to kill the summary only entries until further research is done.

Unfortunately this means losing summary data on days when the cards left out, etc... but at least the data with raw edf files available will be accurate (like it was before)

Anyone have any comments on this?

[Pugsy]

How do we go about killing the summary only files?
I have just been putting the data in version 0.8.8 just to have it. No problems with it.
The double session is not a critical thing for me. I am long past worrying about hours of sleep.
Though I understand that with others this really bugs them.

[jedimark]

I've gone and killed that section in the code.. Now the S9 importer works like the old 0.8.8 version. At least it should be accurate again. (Wahoo, first commit in ages )

It needs a little testing, I'll release an updated windows build after a bit of a thrashing..

Won't be anything really cool to add to it.. just that and the basic F&P Icon stuff for Windows users that is in the latest mac build already.

If anyone is building from source and has a S9, please help me give it a test.. (It will need a purge and reimport.)

A chunk of "summary only" S9 history will go, but this nasty S9 bug should go with it. I wish ResMed engineers would learn to store their summary data properly.

Anyway, after this I'm moving onto a little PRS1 '60 series hacking..

[James1]

james1,

What does this mean? How do you switch what off to restore normality.

Can you please elaborate?

Thanks, Nate

@Nate - see screen captures below of last night's data:

You'll see in the bottom left-hand corner of the first shot where Sleephead lists the sessions that there are two, out of sync by an hour. The first one is the incorrect one (I was actually in bed between 2346 and 0628). You can also see the effect on the flow graph - double length of sleeping, a halved AHI and an apparently clear area for the last hour.

Now have a look at the second shot - you'll see I have switched the incorrect session off by moving its slider (it now shows red in the slider to show it's not being considered in Sleepyhead's calculations). The flow graph, AHI and length of session is now correct.

Hope that helps.

@Mark - will killing the "summary only" files affect the backup of the raw ResMed data? If not - then I'm happy!

[Pugsy]

You'll see in the bottom left-hand corner of the first shot where Sleephead lists the sessions that there are two, out of sync by an hour.

Well...I will be darned. That does work. I my case it is the second session that is incorrect. Click on the bar to make it red and it works. Even fixed my whacko EPAP pressure thing.

[NateS]

james1,

What does this mean? How do you switch what off to restore normality.

Can you please elaborate?

Thanks, Nate

@Nate - see screen captures below of last night's data:

You'll see in the bottom left-hand corner of the first shot where Sleephead lists the sessions that there are two, out of sync by an hour. The first one is the incorrect one (I was actually in bed between 2346 and 0628). You can also see the effect on the flow graph - double length of sleeping, a halved AHI and an apparently clear area for the last hour.

Now have a look at the second shot - you'll see I have switched the incorrect session off by moving its slider (it now shows red in the slider to show it's not being considered in Sleepyhead's calculations). The flow graph, AHI and length of session is now correct.

Hope that helps.

James1, This is absolutely brilliant!
Thank you!
What a relief that the accurate data is still there and not lost!

Now the tricky part of going back to each night since the DST cutoff date on March 11 and figuring out which of the lines to turn off - it gets especially tricky when I may have turned the unit off and then on once in the middle of the night, as I then have to find and pick the right combination of two lines out of the four lines showing. Any suggestions on that?

Nevertheless, this is brilliant!

Regards, Nate

[James1]

James1, This is absolutely brilliant!
Thank you!

No problem Nate!

Have just had a look through my data to see if there are any nights where I had two sessions. I can't find any, sorry. But gut feel is that if, as I have, the first session is the wrong one, then it would be the first, third, fifth etc, or as Pugsy has noted with the second session being incorrect, it would be second, fourth etc etc. One thing I noticed is that is you switch off the wrong session, it gets rid of all data. So depending what happens, you should be able to work out the right one.

Cheers,

James

[Zzzzzzzzzzz...]

James1, This is absolutely brilliant!
Thank you!

No problem Nate!

Have just had a look through my data to see if there are any nights where I had two sessions. I can't find any, sorry. But gut feel is that if, as I have, the first session is the wrong one, then it would be the first, third, fifth etc, or as Pugsy has noted with the second session being incorrect, it would be second, fourth etc etc.

Cheers,

James

That's been the case with my duplicates as well. The first one is wrong and always shows the "No Data" cube when I look for the detailed views on my Mac.

[jedimark]

@Mark - will killing the "summary only" files affect the backup of the raw ResMed data? If not - then I'm happy!

Nope.. It will still backup everything fine.

Just will not import the history for really old data where there is no EDF files, or on days where the card was left out..

A pain, but it should stop the double-ups and crud..

[DiverCTHunter]

Probably OT, but John_Jason seemed to be having some problems compiling for RedHat

I've got a copy of VirtualBox and can toss together an image of each of the major distros if you want a clean binary build environment.

[KrisasMan]

Speaking of other compiles.. I wonder how hard it would be to port this over to iOS. I would imagine my experience is similar to most, the clinician has to take the CPAP to a computer in another room and import the data and print it out. With more iPads going into clinics it would be easy for them to get a 'camera kit' from Apple and plug the SD card or USB stick directly into the iPad and read off your data. More interactive too. Obviously this would only work for the recent model units but then this would let you cut out code you don't need for older machines.

To expand on this maybe you could charge $10 (I would totally pay) for a user version and then $$$ for a clinic version that either stores data to a server or (since I suspect they don't keep the data) just wipes itself of the user data automatically.

If nothing else it would provide some income for your great work.