CPAPtalk.com

M.D.Hosehead wrote:

3)
If I'm following correctly, you have the ability to change your algorithm to mark 5 consecutive blocks (10 seconds) as an "event".

I've lowered the block limit to 5 and also slightly increased the RMS threshold so more events will be detected.

You'll have to reupload files (the detection algorithms are applied when the upload is made, not when viewing).

Let me know how the new settings look! Thanks!

What an outstandingly selfless piece of coding, cpapdork.

There may be many here who cannot afford, or get a prescription for the Encore software necessary to take more active control of their therapy.

You have done them a great service.

Hello CPAPDORK, and thanks for creating & sharing the technology!

I've only been using the therapy for three days now, and ordered the Encore Pro yesterday. I find your web based application quite interesting and it was working fine with IE 8.0 and my Dell Mini laptop running Windows 7 Home Pro.

Very interesting information for a newbie!!!!!

Thanks!!!
SW

Southernwind wrote:Hello CPAPDORK, and thanks for creating & sharing the technology!

I've only been using the therapy for three days now, and ordered the Encore Pro yesterday. I find your web based application quite interesting and it was working fine with IE 8.0 and my Dell Mini laptop running Windows 7 Home Pro.

Very interesting information for a newbie!!!!!

Thanks!!!
SW

Have to ask, where did you order Encore Pro?

Greetings cufamtx!
Regarding you question on where I purchased the Encore Pro....

SW[/quote]
Have to ask, where did you order Encore Pro?[/quote]

Sorry for misleading you on the Encore Pro; I just checked the order with CPAP.com, and see that it was the EncoreViewer 2.0 that I ordered.

Still very new at the usage and lingo associated with CPAP.


SW

cpapdork wrote:

M.D.Hosehead wrote:

3)
If I'm following correctly, you have the ability to change your algorithm to mark 5 consecutive blocks (10 seconds) as an "event".

I've lowered the block limit to 5 and also slightly increased the RMS threshold so more events will be detected.

You'll have to reupload files (the detection algorithms are applied when the upload is made, not when viewing).

Let me know how the new settings look! Thanks!

Here's what the last two night show.

Jan 1-2:
EV 2.0 flagged 5 events--one OA, two HY and two CA. DataViewer 0.3 didn't note any "events."

Jan 2-3
EV flagged one HY and three CA's. DataViewer identified one of the CA's as an "event."


At the times of the EV flags, there are spells of no airflow that last 10 or so seconds, but containing a slight movement of air, like a small breath. EV seems to disregard that and identifies a CA, but to DataViewer,'s algorithm, the quiet period doesn't last 10 consecutive seconds.

It's still valuable to be able to look at the waveform.

Checking in to let you know I'm continuing to progress steadily on the web service for viewing PR data. I had hoped to release an update this weekend but I wasn't able to code as quickly as I planned.

The next update will include though:

• improved multiple-file upload interface that will support (and require) upload of 001/002/005 files.
• basic settings display on the report, including APAP/CPAP mode, pressure min/max, and flex type/amount
• a "table of contents" page to browse your past uploaded sessions

Thanks for everyone's input so far, especially those who have emailed me about providing extra data. It has been extremely helpful in deducing the contents of these files. With the next release, I will be able to capture a great deal more data from more machines, and so I hope I will be able to continue extracting more info.

By the way, if there are any coders out there interested in helping, I've put the code project over on sourceforge, and you're very welcome to help out! http://sourceforge.net/projects/onkor

M.D.Hosehead wrote:
At the times of the EV flags, there are spells of no airflow that last 10 or so seconds, but containing a slight movement of air, like a small breath. EV seems to disregard that and identifies a CA, but to DataViewer,'s algorithm, the quiet period doesn't last 10 consecutive seconds.

It's still valuable to be able to look at the waveform.

Is that 'small movement' not the puff of air your machine uses to check your airway?

snnnark wrote:

M.D.Hosehead wrote:
At the times of the EV flags, there are spells of no airflow that last 10 or so seconds, but containing a slight movement of air, like a small breath. EV seems to disregard that and identifies a CA, but to DataViewer,'s algorithm, the quiet period doesn't last 10 consecutive seconds.

It's still valuable to be able to look at the waveform.

Is that 'small movement' not the puff of air your machine uses to check your airway?

Sorry I wasn't clear.

The test puffs sent by the PRS1 are regular and tiny.

What I meant by "slight movement of air" are single waves about the 1/4 the amplitude of a breath. It seems like EV2, disregards those little quarter-breaths" and, if the whole "event" lasts 10 seconds, flags that event as CA.

But it looks like cpapdork's app does not identify these "events" CA's because, by his algorithm, that little quarter-breath breaks up the apnea into two apneas, each one less than 10 seconds.

M.D.Hosehead wrote:

snnnark wrote:
Is that 'small movement' not the puff of air your machine uses to check your airway?

Sorry I wasn't clear.

The test puffs sent by the PRS1 are regular and tiny.

What I meant by "slight movement of air" are single waves about the 1/4 the amplitude of a breath. It seems like EV2, disregards those little quarter-breaths" and, if the whole "event" lasts 10 seconds, flags that event as CA.

But it looks like cpapdork's app does not identify these "events" CA's because, by his algorithm, that little quarter-breath breaks up the apnea into two apneas, each one less than 10 seconds.

Interesting! Is the breath in or out?

snnnark wrote:
Interesting! Is the breath in or out?

Flagged as CA by EV@. Not identified by DataViewer.

I wonder how the machine software calculates this as a CA. Perhaps it figures... The airway is clear, and the tidal volume is less than Xcm so this is a central.

Maybe Cpapdork can do something similar?

Or maybe something as simple as a minimum threshold tidal volume for a "breath"?

Just want to stress that my algorithm for detecting events is extremely basic at this point. It should help you pinpoint "quiet" areas in the waveform but that's about it.

My priority (re event detection) is to decode the recorded of events from the files directly. It will always be valuable to know what the machine itself has deemed to be an event.

Then, I would also like to be able to incorporate other researchers' event detection algorithms to this data, so that users can choose which model they would like to work with. While I believe the machine no doubt does an excellent job detecting various events, there is almost certainly new research and methods out there that could be used to improve detection.

But I likely will not spend much time on improving my own event detection algorithm because a) it would be at the expense of time used to decode the files which I feel is more urgent and b) I am entirely lacking in such expertise

I hope that others who are interested can help out in this regard though. If anyone out there is interested, you can sign up as a contributor at the project page at http://sourceforge.net/projects/onkor/

Thanks again to all who have been contributing their time and thoughts on this board and by submitting their test data.

Hey cpapdork,

I just want to let you know that I just completed my 2nd night of CPAP therapy, and your data viewer has given me a much better understanding of what this sleep apnea is all about. It's much more complex and overwhelming that I thought at first. As I look at the the waveforms I can see that there's a lot going on, but it's the pauses (or events) that I'm beginning to understand a little better. Of course, I get the 7 day average of AHI, but I wanted to know about the duration, as well as clustering, of those events.

Since I had an AHI of 58.8 during my titration, I was pretty excited to see it come down to 11.9 after just my second night. Unfortunately, most of them were clustered within a 31 minute window last night. Several of them were between 40 and 52 seconds. At this point, I'm not even sure if events of that duration are that unusual. At any rate, I'm going to try to get ahold of the the res therapist to see if I can get some answers....

Thanks again...