Advice Needed-Is my machine going nuts?

NateS wrote:

Papit wrote:

myconoclast wrote:, . . . but I really think there is something wrong with our actual machines and not the software. . . .

Hi again, Nate. Been following your thread. I agree with myconoclast and would guess something intermittent is acting up in both your machines. I'd contact the mfr directly and inquire. Good luck to both.

Hi Papit!

I understand the logic of what you are suggesting, but how am I going to get past them saying that our ResScan software is supposed to be available only to registered professional clinicians and what are you doing with it and how did you get it? And we provide support of our software only to those who are authorized to obtain it? What would Microsoft say if I called them for support for Excel or Windows for example and told them I was having trouble with a copy I got from a friend or a street vendor in Singapore? {I don't do that - I buy all my regular computer software from authorized vendors.}

And also I would hate to be responsible for causing ResMed to track down and block off all members/patient's access to the ResScan software.

Regards, Nate

_________________

My guess in response to your questions, Nate:
1. Re. patient use of ResScan (ditto for Encore) software. If asked, I'd tell them the truth, but keep it general. 'I found an online source.' I doubt that they will raise such questions in this day and age of common public use of technology and the web.
2. Re. software support. I would be asking the mfr about concerns over the machine's (not the software) performance and reliability after only 15 months of use; concerns arising from dramatically inconsistent and missing data types from one day to the next. After all, it is the machine that records your breathing data and its performance parameters on the SD card.
3. Re. mfr tracking down patients' use of software. Ain't gonna happen, Nate. Makes no sense (to me anyway). I'm sure they're already well aware of the availability of their software and use by many astute patients. It would be a public relations mistake for them to attempt to inhibit our use of it. And, if even partially successful, it would jeopardize some futiure purchases by aware and savvy patients.
So I wouldn't be afraid to talk to them. You have a legitamate concern about your machine and, if they aren't already aware of the problem that more than one of us is seeing, I'm sure they will be most interested in looking into this. Your machine mfr is a known high quality leading manufacturer. I would expect them to have some good quidance on the subject of your concerns.

NateS wrote:but how am I going to get past them saying that our ResScan software is supposed to be available only to registered professional clinicians and what are you doing with it and how did you get it?

They know we use it. They know where we get it also! Did you happen to miss these threads?

viewtopic.php?f=1&t=87086&st=0&sk=t&sd=a

viewtopic.php?f=1&t=87048&st=0&sk=t&sd=a

idamtnboy wrote:

NateS wrote:but how am I going to get past them saying that our ResScan software is supposed to be available only to registered professional clinicians and what are you doing with it and how did you get it?

They know we use it. They know where we get it also! Did you happen to miss these threads?

viewtopic.php?f=1&t=87086&st=0&sk=t&sd=a

viewtopic.php?f=1&t=87048&st=0&sk=t&sd=a

Thanks for that tip - I missed that second thread completely!

Regards, Nate

I examined the source files from my SD card using an EDF viewing program, and the crazy spikes in the graphs are right there in the source files. For each session, there are 3 graph files created (for the s9 AutoSet at least)- the BRP file which contains Flow Rate and Mask Pressure, the SAD file which contains Pulse and SpO2 (if applicable), and the PLD file which contains Mask Pressure, Therapy Pressure, Exhale Pressure, Leak, Respiratory Rate, Flow Limitation, Minute Vent, Snore Index, and Tidal Volume. There are no graphs for AHI and Events that I can see; that info must be extrapolated from the other graphs.

The way I see it, our machines are malfunctioning somehow when they record the data. I think ResScan realizes there is something wrong with the data and doesn't show it. Sometimes the malfunction just affects the BRP file, in which case Flow Rate is missing in ResScan but all the other graphs are there, and other times it affects the PLD file as well, in which case all the graphs are missing in ResScan other than AHI and Events. SleepyHead goes ahead and shows the crazy data, giving us the graphs that don't make sense.

I am pretty sure the statistics reported are good, but it is concerning that the machines are malfunctioning this way, especially because our machines aren't very old. I will try contacting ResMed (I don't have to mention ResScan, just talk about the huge spikes in the data that are viewable using an EDF viewing program.)

myconoclast wrote:I examined the source files from my SD card using an EDF viewing program, and the crazy spikes in the graphs are right there in the source files. For each session, there are 3 graph files created (for the s9 AutoSet at least)- the BRP file which contains Flow Rate and Mask Pressure, the SAD file which contains Pulse and SpO2 (if applicable), and the PLD file which contains Mask Pressure, Therapy Pressure, Exhale Pressure, Leak, Respiratory Rate, Flow Limitation, Minute Vent, Snore Index, and Tidal Volume. There are no graphs for AHI and Events that I can see; that info must be extrapolated from the other graphs.

The way I see it, our machines are malfunctioning somehow when they record the data. I think ResScan realizes there is something wrong with the data and doesn't show it. Sometimes the malfunction just affects the BRP file, in which case Flow Rate is missing in ResScan but all the other graphs are there, and other times it affects the PLD file as well, in which case all the graphs are missing in ResScan other than AHI and Events. SleepyHead goes ahead and shows the crazy data, giving us the graphs that don't make sense.

I am pretty sure the statistics reported are good, but it is concerning that the machines are malfunctioning this way, especially because our machines aren't very old. I will try contacting ResMed (I don't have to mention ResScan, just talk about the huge spikes in the data that are viewable using an EDF viewing program.)

Nate, I see the same thing in the one BRP file of yours I looked at with EDFBrowser. I'll look a little more later, but right now I'd say your machine is going nuts. The spikes are only one data number long which is 1/25 of a second. Disregarding the spikes the trace looks normal. I think Resscan may be ignoring that file because it doesn't like the spikes, or the spike numbers are mucking up the CRC calculations. Resscan checks the CRC and if it isn't correct it considers the data to be corrupt or bogus and won't graph it. SH doesn't check the CRC because none of us, including Mark, knows how it is calculated. That's why those graphs show in SH, but not Resscan.

If a particular graph is not viewable in Resscan for a given day, the option to view it also is not shown. That's why your option box is missing those choices.

myconoclast, the AHI and events data come from the STR file in the root directory of the card. SH doesn't use the STR file. I don't know why Mark doesn't. He calculates the summary data directly from the detail data. I wouldn't trust the statistics on the Summary page at this point. But they may be good if the screwed up data is being screwed up during the card write process and not during the calculation process. That's what I'm inclined to think is happening. Something like dirty power during the write process may be causing the bad data. Or the circuitry that does the writing is experiencing erratic signal processing.

In any case I think both of you have bum machines that probably are going to continue to deteriorate. You might wait a week or two and see if the problem get worse. If you're lucky the machine will just plain go bad and quit writing the SD card altogether. It'll make getting them replaced less of a hassle.

myconoclast wrote:I examined the source files from my SD card using an EDF viewing program, and the crazy spikes in the graphs are right there in the source files.… I will try contacting ResMed (I don't have to mention ResScan, just talk about the huge spikes in the data that are viewable using an EDF viewing program.)

idamtnboy wrote: Nate, I see the same thing in the one BRP file of yours I looked at with EDFBrowser. I'll look a little more later, but right now I'd say your machine is going nuts.… In any case I think both of you have bum machines that probably are going to continue to deteriorate. You might wait a week or two and see if the problem get worse. If you're lucky the machine will just plain go bad and quit writing the SD card altogether.

The plot thickens! Thank you very very much myconoclast and idamtnboy for your expertise and input.

I did get good graphing the last two days, but I will wait for the other shoe to drop and am closely following this thread!

Regards, Nate

Nate,
Here are three graphs from EDF Browser. These are from the BRP file for the night of March 10. Excuse the large size of the graphs. When I tried resizing them down detail was lost. You just have to zoom out your browser window to see the entire graph. In Firefox thats done with Ctrl-+ and Ctrl--. Ctrl-0 takes you back to default window size. I'm writing this in large font so you can read it easier when you have your window zoomed out.

This first one shows a randomly selected portion of normal flow data for 20 seconds. You can easily see the 1/25 second data points.



This one is for 10 seconds and shows one of the times where you have bad flow data. As you can see the graph points vary widely and wildly, as much as 60 L/sec, which equals 3600 L/min.



This is a portion of the mask pressure data for 10 seconds. By default the mask pressure data is not shown in Resscan, and maybe not SH. You can make a change in the XML file that controls Resscan to show it, but I've concluded it really doesn't help to understand the therapy. What you see here is a very, very, clear example of the machine putting out bad numbers randomly. They are intermixed with good numbers and show up as single data point anomalies in an otherwise smooth curve.



Unless you need for me to, I don't plan to play around any more with your data. There's plenty here to substantiate your machine has gone, or a least is going, cuckoo.

Thanks so much for you time, effort and analysis, idamtnboy.

I really appreciate it.

Best wishes, Nate