S9 flow skew bug... Anyone else seeing this?

The data from last night are skewed by about a minute.
I'd seen the beginning and end of the data skewed before by about 30 seconds, but now the data itself are entirely skewed.
Comments?

I haven't had that...
I have sessions and flow data that did not show up.
After closing the patient, exiting ResScan and ejecting the card... it showed up upon re-opening.
Just too much info stored in the memory... I guess.

I presume it was brand new info that you had just DL'ed.

Hope that works for you... nobody wants bugs...

Nord

Interesting, I've not hit upon this yet. I have noticed though that if you open up the patient record, then click DOWNLOAD DATA and import the data from SD card, once complete open up last nights record, but before clicking SAVE eject the SD card suddenly you lose the FLOW and other high resolution data.

It seems like when you import the data from SD card it is still using it from the SD card and not from the hard disk.

dave21 wrote:Interesting, I've not hit upon this yet. I have noticed though that if you open up the patient record, then click DOWNLOAD DATA and import the data from SD card, once complete open up last nights record, but before clicking SAVE eject the SD card suddenly you lose the FLOW and other high resolution data.

It seems like when you import the data from SD card it is still using it from the SD card and not from the hard disk.

That's really weird!

Here's how I fixed my skewed data, at least from here on out:

To check your skew period:
Put your event graph above your flow graph for both view panes. Choose whichever 'time window' best suits for the upper 'navigation' pane, and '5 minutes' for the lower 'detail' pane. Click on one of the most obvious apneas (the longer the better) in the upper pane to get close, then center it as you please in the lower pane. You'll likely see that the event as marked on the lower pane is skewed forward a certain period of time from the identical data in the flow graph in your detail pane.

Take a snapshot of your ResScan 3.10 window and bring it into your favorite image editor. With the rectangle tool, draw a rectangle whose upper right corner sits square in the 'lollipop' icon of your target apnea in the event and whose bottom left corner sits at what you approximate to be the very end of the apnea as it actually appears in the flow data.

Next, create a similar rectangle as a meter to measure the distance from one 1-minute tick-mark to a consecutive minute tick-mark in the minute ruler for the five-minute detail pane.

Divide the width of your minute rectangle by the length of your skew rectangle to get an approximation of the number of minutes your skew is off. Multiply this number by sixty to get the number of seconds instead. (See below.) The only thing is that the data for the hypops seem to be skewed differently. I wonder if there is more than one skew bug here... Either that or I still don't really 'get' how to identify a hypopnea on a graph.



To reset your card and eliminate skew:
I've saved a copy of my sd card from last night and then wiped the card using ResScan. In doing so, the card only contained settings data only. I put the card into my machine and it accepted the wiped card, but did not wipe all the accrued summary data still on the machine. I locked the card before putting it into the usb card reader on my computer, then downloaded and discarded duplicate data. All of my data for last night (and the nap too) are now lining up correctly as shown below. Data for the previous skewed nights remain skewed - not much to be done about that. I would amend best practices for card use thus:


Amended Best Practices:

• Open ResScan - allow it to finish opening.
• Lock your SD card.
• Stick the SD card in the reader and allow ResScan to identify the card with the 'Quick Start' feature.
• Select your patient ID.
• 'View Data'.
• 'Discard Duplicate Data'. (I know I've said it before... Flow data erasure is not something that should be so easily accomplished. There should be a check box that allows a default setting the first time. If you are about to erase data from your patient record, you should be asked whether you really want to delete it. It should also be a default setting in tools -> options. I hope ResMed is reading... )
• Read your data.
• Save. (If you're viewing nap data, you could choose not to save here, and wait until your night has finished to save it for real.)
• From here, since the card is locked, it probably doesn't matter if you close ResScan first. It can't change your card.
• Remember to unlock your card when you put it back into the machine.

There also needs to be a more obvious method for wiping the SD card. 'Tools' -> 'Device Settings' -> 'scroll around until you find S9' -> 'erase all data?' is not an intuitive way to wipe the card. How about 'Tools' -> 'Wipe and Reset Device Settings on Card'?

Yes, I have the skew bug too. viewtopic/t50689/S9--Unknown-Apneas.html

Until I read this thread, I didn't even know the SD card had a lock. Thanks for such a clear explanation on how to fix it, DreamDiver.

Edited to add: I had already downloaded my data this morning my usual way, and just now examined it in close detail. There is no skewed data today, Flow lines up with apneas, as expected. Could this be an intermittent problem.

I hope that you "S9 Pioneers" figure out what's going on with the skewed data.

The problem with skewed data is obvious ......

SLEEP APNEA IS CONTAGIOUS .....

your machine has caught your sleep apnea so now it is taking little worktime naps and so it wakes up and has to catch up.

Just as when we have an apnea we don't breathe for a bit ... the machine has an apnea and fails to recognize OUR apnea / hypopnea while it takes a little " nap ". Then it wakes up and says " oh --- you had an apnea I better plant a flag here " . Only it's just a little bit late.

So far, my S9 doesn't seem to have caught my SA. At least I haven't seen the problem yet. Maybe I wasn't looking close enough though. But since the subject came up I have given the graph a small eyeballing.

I did have to wipe my card this morning. A few days ago when I put the card in the computer it was popping up saying there was a problem did I want to scan and fix it. I said now and continued but it took a couple of tries to get Rescan to read the data. This morning though it refused to read it no matter what I did. I checked and the data was written to the card but Rescan wouldn't read. So I copied the last couple of days to the datalog directory ( had already copied most of it there 2 days ago as a precaution ) and then formatted the card then put it into the S9. The S9 did not ask me if I wanted to format. It just said < reading the card >. After a few seconds I took the card out and put it back in the computer and copied the files in the datalog directory on the computer back into the datalog directory on the card and then it worked just fine.

I've learned you need to eject the card either by right clicking on it and using the menu or using the Safely Remove Hardware gizmatchie on the task bar.

But also I wasn't locking the card. Doing that should make it impossible for the files to get messed up by just removing it from the computer I think. I'd bet money I'll forget to unlock it though, and then miss a nights data and have to go buy another one after I stomp on it for a few minutes in frustration at my own stupidity.

I have the skew bug too... it is not only the lining up of the events and flow but also of the pressure and flow... my flow starts about 4 1/2 minutes before my pressure starts. I'll try the card reset tonight and see if it fixes it, but I suspect it will creep back in... especially since it sounds like the problem grows over time.

jmelby wrote:... I'll try the card reset tonight and see if it fixes it, but I suspect it will creep back in... especially since it sounds like the problem grows over time.

I'm wondering the same thing. If we keep our cards locked during data transfer and we still see data skew, we know the problem is with the firmware and not with the computer software interacting with the card. Here's to interesting data.

Arizona-Willie wrote:The problem with skewed data is obvious ......

SLEEP APNEA IS CONTAGIOUS .....

your machine has caught your sleep apnea so now it is taking little worktime naps and so it wakes up and has to catch up.

Just as when we have an apnea we don't breathe for a bit ... the machine has an apnea and fails to recognize OUR apnea / hypopnea while it takes a little " nap ". Then it wakes up and says " oh --- you had an apnea I better plant a flag here " . Only it's just a little bit late.

Good night

I think locking the card from now on might be a better move. It also protects us from accidentally damaging the card, e.g. MOVE vs COPY.

Thanks
Dave

DreamDiver wrote:To reset your card and eliminate skew:
I've saved a copy of my sd card from last night and then wiped the card using ResScan. In doing so, the card only contained settings data only. I put the card into my machine and it accepted the wiped card, but did not wipe all the accrued summary data still on the machine. I locked the card before putting it into the usb card reader on my computer, then downloaded and discarded duplicate data. All of my data for last night (and the nap too) are now lining up correctly as shown below. Data for the previous skewed nights remain skewed - not much to be done about that. I would amend best practices for card use thus:

I tried this approach but even though the card was write enabled, ResScan said it erased it but when I checked I still had all the high res data in the Datalog folder.

In the end I had to format the card and place back inside the S9 and let it write out a new summary table. Thankfully I have my previous card images backed up.

Thanks
Dave

Instead of resetting my card last night, I instead decided to try restarting the S9 itself (unplug and plug back in) to confirm if this is a firmware bug. The skew from last night's data is reduced (it still seemed to be about 30 seconds)... I'm looking at the start of the flow graph vs. start of pressure to determine this. Prior to last night, it was skewed by about 4 1/2 minutes. I think this suggests a firmware bug in the S9.

Looks like we each have different things happening. I'm sure puzzled.

Last nights data was not skewed for me. I suppose time will tell. I was doing so well until the second mask event...

The skew has arrived... but only for the start and end of Flow. All my events are in the correct position for Flow... odd. My starts and finishes are out by about 10-20 seconds depending. Going to do some more investigatin'

Nord