SleepyHead 0.9.8-1 Testing Builds

bwexler wrote:Mark, I may have found another bug.
When I dowload my data it defaults to March 2090. When I atempt to type in more appropriate date at the bottom of the overview page it does not accept what I type.
For example I delete the first 2 digits in the custom from field and type <1><0>the <1> goes in the month field and <0> goes in the day field.

In the daily view I get the same default. Yesterday I scrolled through 75 years to the correct month. that worked. today I highlighted the year (2090) and typed in 2014 and that worked better.
I am not sure my current machine will last me untill 2090.

In 0.9.6 I had the problem with missing days when I forgot to put the SD card back in my PRS1 960. I was trying to check for that error when I found the one above.

I am running Windows 7 Pro Service Pak 1 on Home brew AMD FX 6200 6 Core 3.8 Ghz 16 GB RAM SH 0.9.8-1 64 Bit Open GL.

That's bizarre..

On PRS1, the dates are taken from the timestamps inside the block headers of the .001/.002/.005 files.
These timestamps are stored in very standard format, which is seconds since UNIX epoch, which was January 1st, 1970, 00:00am UTC, and pretty they are much impossible to screw up.
The conversions are done by Qt QDateTime function calls.

If you are importing and getting the wrong date, it's very likely you have a completely corrupted .00x file screwing things up on your devices SDcard. (The header bytes, which include the timestamps are protected by an additive checksum, and sessions that don't match this are dropped. There is a secondary 16bit checksum test, but it's switched off in the interests of import performance, as the likelyhood of this happening is very VERY low, i'll make sure to reenable this as a preference to make sure faulty sessions that pass the first checksum test get dropped.)

There is a slim possibility you have a corrupt value set to the clock drift in preferences...

If possible, I'd like to see a .zip (or .7z!) of your SDcard to try and replicate exactly what's going on here.

/Mark

Mark I would be happy to send you the 7Z file. It is just under 40 MB how and where should I send it?

Is it possible that my CMS 50i has caused the corrupted date?
If so what is the best way to send you that data? I don't know how to get it out of the watch. do you save it somewhere in Sleepyhead?

I did find that a CMS 50E import led to corrupted dates in Sleephead 0.9.3. I never did get it straight. Eventually it disappeared into the murky past and I ignored it. I know, different CMS and different SH build but your remark triggered the memory.

Hey Mark. Thanks so much for all your work. The first thing I noticed when I checked out the new preferences was the warning about Resmed session splitting. When I follow the link though, there's nothing in the wiki entry. What are the pitfalls of changing the time that the session splits?

JQLewis wrote:Hey Mark. Thanks so much for all your work. The first thing I noticed when I checked out the new preferences was the warning about Resmed session splitting. When I follow the link though, there's nothing in the wiki entry. What are the pitfalls of changing the time that the session splits?

Oops.. I forgot to edit the wiki page

Hovering over the tooltip Don't Split Summary data, this message pops shows up, which explains some of the reasoning behind ResMed session splitting issues.

This setting should be used with caution... Switching it off comes with consequences involving accuracy of summary only days, as certain calculations only work properly provided summary only sessions that came from individual day records are kept together.

ResMed users: Just because it seems natural to you and I that the 12 noon session restart should be in the previous day, does not mean ResMed's data agrees with us. The STF.edf summary index format has serious weaknesses that make doing this not a good idea.

This option exists to pacify those who don't care and want to see this "fixed" no matter the costs, but know it comes with a cost. If you keep your SD card in every night, and import at least once a week, and you won't see problems with this very often.

Changing the noon split time is not a good idea on resmed. I left it so SleepyHead is able to do it, but it will cause data accuracy problems that I have no way of working around, but only with summary-only sessions.

ResMed machines were simply not designed with shift workers or late sleepers in mind..

The new AirSense 10 machines improves this situation a little, as it doesn't chomp your data every 7 or 30 days.. so provided you leave the card in every night, you won't see this problem anywhere near as much.

For the S9.. if you don't want to see summary day split problems, be a little pedantic about making sure the SDCard is in the machine, and make sure to import at least once a week!

bwexler wrote:Is it possible that my CMS 50i has caused the corrupted date?

The CMS50i (and the new CMS50F's) are the first Contecs CMS50's that bother to provide a clock with date.. However if you are using the time from the oximeters clock and it's set wrong, this could indeed screw up.

If Oximetery is causing the problem, you will notice only have SPO2, Pulse and Perfusion Index charts on that day... Which you should be able to delete the oximeter session from the advanced menu and have the problem go away next time you restart.

SleepyHead has an option (enabled by default) to set the Oximeters clock during import from the PC's time.. (If the PC clock is somehow set wrong, it could screw up too, but most computers sync to an internet time server now)

Personally, I find it more accurate to sync to CPAP sessions... And it's quite easy if you have your CPAP machine set to auto start:
All I do is go into the oximeters menu, go to the start recording option, but don't click start yet. I then mask up and hold my hand near the CPAP vent so I can feel once the air starts.. Once I feel the machine start up, I click start on the oximeter.

The next day after I import, I select the radio button to sync from cpap, select the date, and choose the CPAP session that I started the oximeter at.
(If you go to the date first, there's a known bug where you have to go off the calender day and back on it to find the sessions..)

It is definitly the CMS 50i causing the problem with the date.

First, I am new to sleepyhead. My previous machine was not supported so when I got my S9 it was my first opportunity to use it.

I do a lot of two sleeps in a day, most of the time. Since Friday, I've been battling something, not quite a cold but with lots of achy so I've been drinking lots of fluids and trying to get lots of sleep. Been successful so far and it looks like whatever it was is going away. I have not had any trouble sleeping.

Looking at the reports can be a bit confusing though. Examples over the past couple days (just the left side looking at the sessions):

Oct 4th, 5th, 6th and 7th:



What is confusing is what is actually being reported.

On the 4th, the total at the top is 1h 43min, But at the bottom it shows 3 sessions of 1h 43m, 4h 47m and 5h 3m (for a total of little more than 11hrs)

On the 5th it shows 11h 29m on the top and 4 sessions of 3h 13m, 1m, 4h 40m and 3h 35m which matches the top.

On the 6th it shows 7h 6m on the top and two sessions of 4h 13m and 2h 53m which appears to match the top

and on the 7th it shows 3h 21m with one session of the same.

On some of the reports the bottom has a golden tone for a session, a blue tone and sometimes a white background.

What do the different color tones mean and is that related to what is displayed on the right? It seems like I can click on the session bars at the bottom and I can get some really weird numbers on the right with AHI's like 30.28 etc.. gets a bit confusing.

Followup on the above... sometimes when I go to a day, I get completely different numbers like below which is for Oct 4th. This time is shows 9h 50m with the same 3 sessions listed before. AHI is 30.28

Krelvin wrote:Followup on the above... sometimes when I go to a day, I get completely different numbers like below which is for Oct 4th. This time is shows 9h 50m with the same 3 sessions listed before. AHI is 30.28

when you're clicking on the sessions at the bottom of the detail page, you're turning them on and off, and so only seeing parts of the day.

it's confusing because there's no explanation of that fact.

Krelvin wrote:Followup on the above... sometimes when I go to a day, I get completely different numbers like below which is for Oct 4th. This time is shows 9h 50m with the same 3 sessions listed before. AHI is 30.28

But you are only looking at two of the three sessions. Under "Session information," the first session block shown is clear, indicating it's disabled. In addition, under "CPAP sessions" you can see that the first entry shows a red block instead of a green one like the others. That's another sign that the first session is ignored. If you look at the times under the graphs, you'll see (I think) that the time starts with 21:01, and that only two sessions are shown with a break between them.

You can turn the first session back on by clicking either on the white block or on the red indicator. This ability to turn sessions on and off is very useful to me in looking at the efficacy of therapy in individual sleep sessions during the night, particularly if there's a bad one mixed in with good ones.

As Palerider already said, it's confusing because it's not explained anywhere. It sure confused me before I figured it out! Now it's one of my favorite features.

Okay that makes more sense. Clicking on the Red or green seems to work better than clicking on the blocks. I had been clicking on the blocks before but never could figure out what was actually happening between the yellow, blue and white backgrounds.

Krelvin wrote: had been clicking on the blocks before

Goes to show you how unobservant I am...I didn't even notice that the boxes were there.
I have no idea when those were added or even if they had been there all this time.

I would always go to the green/black thingy to turn on or off the sessions.
We had a little bug a long time ago (fixed a long time ago too) where we would have to sometimes turn off a session because the total hours slept and sessions were buggy.

The help page still says:

Currenly supported machines:

CPAP
Philips Respironics System One (CPAP Pro, Auto, BiPAP & ASV models)
ResMed S9 models (CPAP, Auto, VPAP)
DeVilbiss Intellipap (Auto)
Fisher & Paykel ICON (CPAP, Auto)

It does support the ResMed A10, right?

cathyf wrote:It does support the ResMed A10, right?

read this: viewtopic.php?f=1&t=101108&st=0&sk=t&sd=a#p943672