[Renamed] S9 graph display problem

Imagine my near horror this morning when I looked at the graphs for last night's data and saw about a 40+ minute skew between the event graph and flow graph. After spending a big part of the day working to figure out what's going on here's what I've come up with. It is primarily a Resscan s/w problem, not the S9 machine or the SD card.

I did two things basically. I extracted the data files from the EDF files and looked at them with Lotus 1-2-3, and played around with the window time frame in the detail graphs in Resscan. Now, I looked in detail at only the first event of the night, a central, so it's possible, but doubtful, that there could be some differences in recording obstructive, hypopnea, and unknown events.

First, the S9 records in the EVE file the onset time of an event as the time after the first real breath after the event has ended. This makes sense because the algorithm needs to analyze data for a period of time to determine whether or not a period of non-flow exceeds the 10 second threshold to be considered an apnea. When this event is graphed in Resscan it looks like it is placed in the timeline right at the end of the non-flow period, i.e., just before the first real breath. You can see this in the charts below.

Second, all of the data I looked at among the various EDF files is consistent, and none of that data corresponds to the wrong timeline shown in Resscan charts.

Third, the correlation of the upper and lower charts jumps around in relation to the time frame being used for those charts. In the first graph below the top chart is shown at 8 hrs, the lower chart at 10 min, and the event and non-flow vary by 40+ minutes. As the time frames of the upper and lower graphs are made shorter and brought closer and closer together the event display comes closer and closer to the end of non-flow period.

The last graph is a graph from Lotus showing flow vs time of day. The time of day on the chart is derived by adding the start time from the header part of the BRF.edf file and the time, in increments of 1/25 sec, listed in the data part of the edf file. The time for onset for the event is recorded in the annotation portion of the EVE.edf file and is exactly the same as the time in the BRF file at the end of the first breath after the non-flow period.

I'm convinced the problems that were first discussed here back in April is a Resscan software problem, not an S9 machine or SD card problem. Why that hasn't been fixed now in version 3.12 is a good question. Resmed has dropped the ball in my opinion.

Have you tried to force a screen refresh through the graphing options panel?

ozij wrote:Have you tried to force a screen refresh through the graphing options panel?

Yeah, didn't help. Do you close the options window and then open it up again, or just deselect & select without closing/reopening?

idamtnboy wrote:

ozij wrote:Have you tried to force a screen refresh through the graphing options panel?

Yeah, didn't help. Do you close the options window and then open it up again, or just deselect & select without closing/reopening?

I deselesct and the chart immedialtely disappears, I reselect and it comes back alingned -- although sometime I had do deselct and reselct anonther chart too one before thing really settled. I don't close and reopen the panel, IIRC, but I've been doing it pretty automatically -- next time I'll try to pay attention to what I do,

EDIT: See below... Everything looks like it corresponds to me. I don't see the problem.

There are two panes: the navigation pane and the detail pain. The navigation pain shows two light-blue vertical lines that define the limits of the detail pane. The red vertical line in the lower pane denotes the center of the area delimited between the two vertical blue bars in the upper pane.

• In the first image, you have it set to a navigation scope of 8 hours. The blue lines in this pane delimit the narrower 10 minute scope shown in the lower pane exactly as I would expect.
• In the second image, you've zeroed in on a two-hour navigation scope and a ten-minute minute detail scope as defined once again correctly by the blue lines in the navigation pane. Again, this is exactly as I would expect.
• In the third image, you're comparing the upper ten-minute scope to another ten minute scope. At this magnification, you may notice that the two panes may not align. That's normal. In fact, if you slide the scrollbar in the lower pane too far in one direction or the other, you'll probably notice that the blue vertical lines are no longer visible in the upper pane.

With both scopes set to ten minutes, you're not really making use of the navigation/detail concept but rather comparing one ten-minute portion of the data to another ten minute portion. They are not meant to be identical views. Just because you've slid the detail pane beyond the scope of the navigation pane does not mean there is a problem. The navigation pane and detail pane may not necessarily line up automatically when you start changing scope.

To show what I mean:

1. Go again to the graphs for the day in question below.
2. Select the navigation pane '8 hours' view.
3. Center the data as you like it in the navigation pane.
4. Put the detail pane at 10 minutes.
5. In the nav pane, click once directly on any apnea or hypopnea that interests you.
   You'll see the blue scope lines jump to center themselves around where you clicked in the pane.
   The lower detail pane will jump to the ten minute scope you centered your mouse-click on in the nav pane.
   The upper pane will not move, but the bottom pane will have the event you selected somewhat centered on the event you wanted to explore.
6. Now, click the left-hand or right-hand arrow icon in the detail pane's horizontal scroll bar.
   Notice that as you slowly scroll the data in the lower pane, the blue vertical bars in the upper pane slide to accommodate the new view defined in the lower pane.
   As you increasingly narrow the difference in scope between the upper pane and lower pane, the easier it will be for you to slide the blue bars out of view on the nav pane. If you're views are inverse in scope, you probably won't see the blue bars at all.
7. Decrease the navigation pane scope to 2 hours and the detail pane to five minutes and try the same thing.
   In the lower pane, scroll the horizontal scroll bar way past the limits of the two hour scope of the upper pane. You should notice that the blue vertical delimiters are off the page, no longer showing in your upper pane.

The upper pane is oranges and the lower pane is apples, even when they're at the same time scope. Your lower pane does not include the event graph. You can only compare apples to apples. In order to corroborate event time, you'll need to include the event graph in the detail pane, just like you do in the navigation pane.

I don't see any appreciable skew at the scopes you've shown us, nor do I see a problem with the software. The upper pane and lower pane do not necessarily center on each other. All data looks correct, including the data you imported into lotus, but again, you don't have the event data mapped in your lotus graph for apples-to-apples validation.

Does this help, or have I completely missed the boat?

Whoa - wait a second here... The first image. I think I see what you're talking about.
I'm leaving the stuff I wrote above in case it might help others. However...

This is what you're talking about?


So clicking on the event in the upper pane once does not refresh and give you the correct time reference in the lower pane?
How often do you reboot your computer?
Are other 3.12 users also seeing this? If so, I would concur that this particular anomaly is a ResScan idiosyncrasy and not firmware-related.

DreamDive wrote:Are other 3.12 users also seeing this?

I'm not.

I tried a couple of dozen random samples of nights from both my S8 and S9 data, and the bottom graph set aligns correctly with the selected event on the upper graph set for any clicked-on event on all the sampled nights.

Cheers,

Bill

billbolton wrote:

DreamDiver wrote:Are other 3.12 users also seeing this?

I'm not.

I tried a couple of dozen random samples of nights from both my S8 and S9 data, and the bottom graph set aligns correctly with the selected event on the upper graph set for any clicked-on event on all the sampled nights.

Cheers,

Bill

Something tells me idamtnboy may just need to reboot the system. I've seen similar stuff happen in other programs. Invariable a system cold boot fixes the problem.

bump

DreamDiver wrote:Something tells me idamtnboy may just need to reboot the system. I've seen similar stuff happen in other programs. Invariable a system cold boot fixes the problem.

Sadly, not in this case. I shut down my PC every night and cold start every morning. The timelines were correct until Tuesday, 11/9. They're off today still. I even closed out all the graphs in the options dialog and shut down and restarted Resscan then reselected some of the graphs. Still see the same problem. I see there are a number of xml files in the Resscan program folders. How many of them are used each time the program runs I don't know. I wonder if those could be a source of the problem. We have a media presentation program at church that runs on xml scripts and it acts weird every once in awhile. Maybe xml files are prone to go screwy. Just a wild a** theory on my part.

I haven't uninstalled and reinstalled Resscan yet. That's the next step. I installed it with Ashampoo Uninstaller so uninstalling it is supposed to remove all traces of the program from the HD and registry so the reinstall should be totally clean install just like the first time.

idamtnboy wrote:... I haven't uninstalled and reinstalled Resscan yet. That's the next step. I installed it with Ashampoo Uninstaller so uninstalling it is supposed to remove all traces of the program from the HD and registry so the reinstall should be totally clean install just like the first time.

I can see this is truly problematic. It's very possible one or more of your patient files may be corrupt.

Does the below suggestion seem like a reasonable next step?

• Keep the old patient records for historic purposes but don't put them into ResScan right away after you reinstall. Just store them.
• Reformat your SD card and start a new patient record when you get everything re-installed.
• If you don't see the problem in the new patient data set, this may corroborate the theory that your old patient data are corrupt.

I'd like to know your next step will be and what the outcome is. Good luck.

billbolton isn't seeing the anomaly. I'll toss it out there again -- is anyone else seeing this phenomenon or is this anomaly restricted to idamtnboy's ResScan install?

3.12, on Windows 7.0 and on Winxp, -- no problem on my data or the S9 example.

DreamDiver wrote:It's very possible one or more of your patient files may be corrupt.

I'm skeptical about the patient data file being corrupted. That's because when I zoom in the charts to 30 and 10 second time spans the event markers and the flow graph are dead on, within a second or three, according to the time ruler displayed, of what I see in the BRP_data.txt and the BRP.edf file. I'm looking at the BRP_data.txt file with Lotus and the BRP.edf file with EDFBrowser. As I zoom out on the Resscan displays, the time ruler shifts. At a 10 sec wide window the start time of the graph is off only 1 or 2 seconds. At a 24 hr wide window, in both the nav and detail windows, the start time is shown about 100 minutes earlier than actual. If the data were the cause of the problem the time ruler shift would be constant regardless of the window time span.

ozij wrote:3.12, on Windows 7.0 and on Winxp, -- no problem on my data or the S9 example.

So you are saying that regardless of the time span you set the display window to, the event marker, the flow graph at that point, and the start time of the recording are always shown the same according to the time ruler? If you haven't already double check the start time of the graph as shown in windows at both ends of the time span setting.

I just thought of something. When I start Resscan I've been clicking on Open Patient right away. That command does not complete because Resscan's card driver is searching and finding the data card in my card reader, and it obviously has precedence for Resscan's attention. It's only after the Device info dialog has opened and I've closed it, can I open the Patient file. I wonder if that sequence has corrupted something within Resscan? It shouldn't, but who knows?

idamtnboy wrote:I'm skeptical about the patient data file being corrupted. That's because when I zoom in the charts to 30 and 10 second time spans the event markers and the flow graph are dead on, within a second or three, according to the time ruler displayed, of what I see in the BRP_data.txt and the BRP.edf file. I'm looking at the BRP_data.txt file with Lotus and the BRP.edf file with EDFBrowser. As I zoom out on the Resscan displays, the time ruler shifts. At a 10 sec wide window the start time of the graph is off only 1 or 2 seconds. At a 24 hr wide window, in both the nav and detail windows, the start time is shown about 100 minutes earlier than actual. If the data were the cause of the problem the time ruler shift would be constant regardless of the window time span.

It is certainly a puzzler. I recognized the shift reduction with higher magnification that you describe, and the fact that the data are correct in the .edf files. However, ResScan does not store the data in its original file formats. (Hence the likelihood that the problem is restricted to something dealing with ResScan and not the flow generator.)

Questions that run through my head:

• Have you done your Ashampoo clean reinstall of ResScan and created a new patient file?
• If so, do you still see the scope delimiter shift?
• If not, do you see the problem with your newly input data too, or just the data from specific days in question?
• Is this a VMWare install on a Mac? Ubuntu host?
• What Specific Windows OS are you using to run ResScan?
• We just had a time change. I wonder if that had any bearing on your data.

Maybe you can restrict the field of cause. My first instinct is your suggestion to reinstall, coupled with the new patient record idea.

If you've been saving images of your SD Card along the way, you may be able to reconstruct all the patient data into a new patient record.

I've just installed 3.12 and have not as of yet experienced any scope/delimiter skew. Data integrity seems untarnished.