SleepyHead Qt v0.8 alpha

[jedimark]

Someone else try this please.

I tried to use Windows magnifier to zoom in on the SleepyHead 0.8.2 display. The graph area does not show up in the windows magnifier. The left hand panel does show up, but not the graphs.

OpenGL is likely the cause of this. Strange though, because screen capture software works.

[justbreathe]

Mark said that sometimes happens when there are short sessions, maybe for some other reasons. Do you have any short sessions on the days there the numbers don't show? If they all don't show, is there anything you think might be atypical about your data sets? Do the numbers show up if you zoom in to take a closer look at an hour or a few minutes?

This is the case for every date and every session. There were some less than five minute sessions the first few days.

Zoom in our out does not make the text readable.

[jedimark]

I can see the text and numbers on the graphs. I am using Windows Vista Home Premium. Any ideas?

What is your system locale? I've noticed this on platforms that aren't english.

I just recently dropped the built in font, so this may help.. Hopefully I can get a new test build online in the next day or two..

[jedimark]

I think I just set a new world record.

My report shows a max leak rate of 36,000 for Saturday. No wonder we've been having a high pressure system around here. Someone call Guinness.

Hmm.. Max calcs aren't time related.. sounds like you found a parser bug or a funny data file on that day. :-\

Don't suppose you could email me that days set of edf files...?

[Otter]

Welcome back, Mark.

If it's a machine bug theres not much I can do about it.. a while ago I was using floating point for time work. Now use 64bit integers, so could be a loss of precision. But I still think it's a bug in the actual EDF files. :-\

I agree. There's really no question that it is in the EDFs. Awhile back, someone said you'd found a way around it, but I went and tested that before repeating it in a review because it seemed a little too good to be true.

Loss of precision from using 64 bit integers causing a 20 second offset on the first event of a new session, when the previous session had no skew? I really doubt that. I think someone just took your comment early on about getting the events to sync up to mean you got all events to sync up, even the ones that don't sync. They were thinking skew. You were glad to have figured out the data formats. Simple miscommunication.

I did think of a way to fix the skewed data, but I don't think it's even worth putting on the back burner yet. Apneas are pretty easy to spot, and so it shouldn't be too hard to write a routine to find them. And by searching for the pattern of apneas in the events, you could find the offset, assuming it's the same through each session, and line everything up. But I have a feeling that taking on ResMed's bugs would not be good for your sanity, so I suggest not even thinking about gravy features like this until sometime after SleepyHead 1.0.

Happy coding.

[justbreathe]

What is your system locale? I've noticed this on platforms that aren't english.

I just recently dropped the built in font, so this may help.. Hopefully I can get a new test build online in the next day or two..

I am using US English on Windows Vista Home Premium.

[archangle]

I think I just set a new world record.

My report shows a max leak rate of 36,000 for Saturday. No wonder we've been having a high pressure system around here. Someone call Guinness.

Hmm.. Max calcs aren't time related.. sounds like you found a parser bug or a funny data file on that day. :-\

Don't suppose you could email me that days set of edf files...?

Glad to. Which files are "edf files?" Just everything with that date? From a certain directory?

The whole card only has 8 MB on it, if you just want me to send the whole thing.

And which e-mail address? PM if you don't want to post it.

I have impossible maximum leak rates on 8/11,13, and 14. 8/12 is OK.

[GoodNight Kiwi]

I finally hacked out how to get at the CMS50's internal recording data.

Hello Mark,

This is my first post on this board !

Last week I bought a Contec CMS-P Oximeter - as reviewed here,
viewtopic.php?f=1&t=35416&st=0&sk=t&sd= ... 15#p317262
that I am very pleased with.

Because this device is just a raw oximeter sensor with no LCD readout or such, it streams the 5 byte protocol continuously.

If I run your v0.8.2 version of SleepyHead - it shows the pleth waveform just fine! Thank you very much!

I was looking around the web for more information on the 5 byte protocol and I came across this interesting document.

http://www.2shared.com/document/ACF5L5m ... col_.html?

It doesn't say what it is for, and it says the baudrate is 4800 (rather than 19,200) but it appears to be a detailed breakdown of the Contec 5 byte code.

It nicely explains where the bit 7 of the "pulse rate" is found for example.

Is that of any help to you Mark ?

Thank you again for SleepyHead so far - it is a great visual tool to look at the photoplethysmograph waveform.

Regards, Geoff (Christchurch, New Zealand).

[jedimark]

I finally hacked out how to get at the CMS50's internal recording data.

Hello Mark,

This is my first post on this board !

Last week I bought a Contec CMS-P Oximeter - as reviewed here,
viewtopic.php?f=1&t=35416&st=0&sk=t&sd= ... 15#p317262
that I am very pleased with.

Because this device is just a raw oximeter sensor with no LCD readout or such, it streams the 5 byte protocol continuously.

If I run your v0.8.2 version of SleepyHead - it shows the pleth waveform just fine! Thank you very much!

I was looking around the web for more information on the 5 byte protocol and I came across this interesting document.

http://www.2shared.com/document/ACF5L5m ... col_.html?

It doesn't say what it is for, and it says the baudrate is 4800 (rather than 19,200) but it appears to be a detailed breakdown of the Contec 5 byte code.

It nicely explains where the bit 7 of the "pulse rate" is found for example.

Is that of any help to you Mark ?

Thank you again for SleepyHead so far - it is a great visual tool to look at the photoplethysmograph waveform.

Regards, Geoff (Christchurch, New Zealand).

Thanks Geoff, I'll have a look at that. It may have something that will improve the current driver.

I'm pretty sure the baud rate is correct for most cms stuff.. if your seeing plethy it's got to be working. The serial hacking tool gave the correct values for the 50e & d+..

I still haven't fired up my dev pc since I got back.. my trusty dell notebook ate yet another power supply. Wife hopefully gets her asus back today so I can have my other notebook back...

[jedimark]

Hi there.. Just letting you know I put an update online to SleepyHead.. v0.8.3

Direct link here to the new binary here.

http://sourceforge.net/projects/sleepyh ... e/download

It's just the exe.. it needs to go in the folder with the DLL's of the old version.. if you haven't got them already grab this too, and copy the new binary over the top..

http://sourceforge.net/projects/sleepyh ... p/download

Fixes a lot of bugs to do with Min/Max/Average/90% calculations..

Preliminary ResMed VPAP stuff is done and needs a testing.. The S9 Oximeter may or may not work.. (It should, but I'm kinda replanning parts of the oximetry backend)
Overview still isn't done.. I want to fix the vertical scrolling bug before moving on to it.. (Well, I want Qt to fix it, it's their bug, but I'm trying to find a work around)

PRS1 users that are building 64bit linux/mac builds may note better flag alignment.. I've been neglecting the 64bit stuff... but now my 32bit dell has eated yet another power supply, so I'm forced to use my big quadcore PC for a while..

[archangle]

If this is the version I just tested, existing PRS1 users need to know that they need to import the root directory of the SD card. i.e. the parent directory of the P Series directory.

In the previous version, you needed to import the P series directory itself.

The new way probably makes makes more sense.

[Otter]

The new method of browsing for import is nice and snappy in xp x64. It does, however, require navigation from the program directory every time.

Unless I'm misremembering, memory usage has improved again. Okay, maybe not. After clicking around for a while it's about the same. Still pretty reasonable, though. 113mB ATM, and VM is 111.

Switching between days is nice and fast, but then it was in the previous two versions too.

Leak rate is in l/min again, which is good. But if I pull the hose loose from my mask, the leak rate goes above 200 l/min. With autoscaling, this squashes the useful data almost to a flat line. To a lesser extent, this a problem with some of the other graphs too. We are going to be able to set the scale and units for individual traces in preferences by the time we get to beta, right?

BTW, I'm pretty sure you'll find the min and max pressure settings in STR.edf. Would make some sense to use these for the default scale on both pressure graphs. How much trouble would it be to retain those values so that someone clicking though their history would always have an appropriate scale? It might make sense to track machine settings anyway. Don't the bilevel machines offer a choice of waveform? If I had one of those, I'd want to know which option was selected.

I like the combined S9 lowres/therapy/exp pressure graph. This really highlights how the waveform spends most of each breath near the lower pressure when EPR is on.

Calendar is now six days wide instead of five. Still can't drag the the border. The width of the left pane appears to be set by the width of the tables at the bottom of the details tab. There's still plenty of white space in the calendar. Let me know if a screenshot will help.

All days now show the time in the event flags. Another bug bites the dust.

I like the inclusion of day of week when the data crosses midnight. Was that there before? That side of the screen looks prettier too. What'd you do, change the fonts?

I'm glad the notes were retained when it purged the data. I only had one, as I've been hesitant to really use this feature in alpha testing. But it's good practice not to mess with this unless you absolutely have to.

While I was poking around, "brain waves" and "sleep stage" in preferences>events caught my attention again. Do these items mean that you have plans to support the Zeo or a similar gizmo?

Well done! You're making good progress. Thanks again for all the work you've put into this thing.

[Otter]

Oh! You got the trace colors going too. I had gamma turned way down, so they all looked pretty much the same except blue and red.

[fiberfan]

Preliminary ResMed VPAP stuff is done and needs a testing..

I did a import from my card with no problems. I looked briefly at Tuesday morning's data and the charts that are in both SleepyHead and ResScan look the same except for the pressure chart. After work this evening I will look closer at the data and get more info on the pressure chart difference.

[Bookbear]

I,too, imported data with the new release. It looks good (thanks for fixing the total time computation!), but the 90% pressure figures are different than with the earlier version. I'm assuming that there was something wrong with the way that number was reported in the earlier version, hence your comment about fixing bugs in the pressures? Some of the numbers were the same but others were substantially different (11.2 vs. 14.0 for example).

And on a happy note, SleepyHead no longer crashes when I close out! WooHoo!

And for a logo... why not put a cpap mask on that sheep?

Thanks again for putting so much effort into this great resource for us!