Detctive story continued: pb discrepancies...

[ozij]

Well, if -SWS is wagging his tail, - see RG viewtopic.php?p=24739#24739 then I must be the one who won't let go of a bone.

Mystery continued:
Maybe its not the post midnight sessions.

If you compare c/a ticks you count on the detail screen, and c/a in the synthesis counts, you'll see they don't concur. You count (and see) many more on the detail screen than are reported on the synthesis screen.
As far as I can see, it is those ticks that exist concurrently in both the apnea row and the apnea with c/a row that are given this strange treatment: they are counted in the ap., dropped in the c/a, and dropped form both c/a and ap. export data
As far as I can count, the number of missing c/a ticks is the difference between the screen apnea report an export apnea report.

So what in what in the world are those concurrent ticks, and why are they added to apnea count in the synthesis screen, and dropped in the export?

Which is the feature and which is the bug?

Glassgal - do you have any centrals?
Janelle - can you take a peek at how consistent your data is?

Does the pb420e have a clinician manual that might shed light on this? The help file certainly doesn't.
O.

[-SWS]

LOL, Ozi! Not so sure what to think of this canine metaphor you and RG have become fond of using. I'm not wagging any tails regarding your recent and current detective pursuit since I haven't done any actual work on either. I'd feel too guilty wagging for my miniscule efforts! Wish I had more time to dig in and help with this!

With that said I'd like to toss another low-time, low-effort (on my part) milk bone out to the real blood hounds of this mystery... How are mixed apneas represented by the way of tick marks, and which apnea indices do they contribute towards? I believe they receive concurrent apnea and CA tick marks to reflect a mixed apnea event, yet they are supposed to be reflected only in the apnea index.

Could the mixed apnea count be missing from the exported apnea index (presumably incorrect), but not missing in the synthesis AI (presumably correct)?

[ozij]

Does a double tick mark mean a mixed apnea?
If it does, then what we are seeing is a decision (arbitrary? norm guided?)by PB to add them to the apnea index, and subtract them form the apnea a/c index - and a bug in exporting the info.

The amazing thing is that the term "mixed apnea" does not even appear in the help file. Furthermore PB does not inform users of that decision. On the other hand, could the double tick be a bug, a fluke, and the export "real"?
Do other machines report "mixed apneas"? Which index do they add them to?

And to think all this started because I just wanted to put my data into "LongTtermCpapTrends" an excel file I downloaded - and now can't find who donated it....

O.

[-SWS]

The 420e's predecessor, the 418P has discrete "mixed apnea" tick marks represented in the SL3 reports. If you look in the SL3 help file regarding the 418 AutoPAP, you'll see that PB defines mixed apnea as:

Mixed Apnea:
"This refers to a respiratory Apnea consisting of periods when CArdiac oscillations occurred, and periods when there were no CArdiac oscillations.".

Bear in mind that the 420e has identical cardiac oscillation detection capabilities as the 418P and thus an identical central apnea specificity rating as the 418P. So like the pneumotach-based 418P, the sequel 420e is also technically capable of detecting these three apnea conditions manifesting in individual sleep events: 1) purely obstructive apneas (no cardiac oscillations present), 2) purely central apneas (cardiac oscillations present throughout the entire apneic sleep event), and 3) mixed apneas (cardiac oscillations present for only part of each apneic sleep event). The 420e continuously monitors for cardiac oscillations and is therefore just as capable of detecting those mixed apnea events as the predecessor 418P is/was.

But how does the 420e log that mixed apnea? Certainly not with the same discrete row of "mixed apnea" tick marks the SL3 software renders for the 418P report. Rather, I speculate that for lack of a discrete "mixed apnea" report, the 420e acknowledges these mixed events with concurrent apnea tick marks. I think by convention (in the real world) mixed apneas are often (but not always) tallied as ordinary apneas versus a discrete mixed-apnea type or versus a purely central apnea.

If so, the hypothesis I mentioned earlier based on your verrry astute tick-mark observations fits nicely: PB simply forgot to carry the mixed apnea count into the AI index during data export.

[neversleeps]

I know this is incredibly immature, but I just gotta ask:

Does anyone else out there who has been following the PB export data error mystery ever feel like screaming, "What the hell are you guys talking about????????"

Or is it just me....

This is one of those subjects where I think I'm finally understanding it all, only to realize that the more I truly do understand, the more I must admit I actually understand less than I did before I thought I didn't understand any of it.

[glassgal]

Hi O & -SWS,

O - I have very few centrals - 0-2/night, generally.

Last night the synthesis report shows 2 apneas, 0 Apnea/CA, 2 Hypopneas & 0 Hypopneas (FL). The SE export for the same time frame shows 1 apnea, 0 apnea/CA, 2 Hypopnea & 0 Hypopnea (FL).

The detailed report, however shows 2 apnea AND 1 central exactly lined up with the second apnea. The hypopnea & h(FL) data match the synthesis and the SE export.

I think that you are onto something here O.

I will try to find my clinician manual tomorrow. I think that I have one, but it will take some digging. My recollection of trying to read it when the unit was new is that it is heavy going and I am too tired to stay awake to read it tonight. The documentation that I did read was very obscure and hard to follow. I hope that the confirmation of what you figured out helps.

What a learning curve on this!

Thanks for all your time to help us figure this out Ozij & -SWS! You are both the best!

[ozij]

Gang - I found this great picture, and it won't show when I paste the URL- so click: Self Portrait?

So... I went to the 418 demo data (file 2312) and discovered that:
In the synthesis screen for 418 data, the mixed apneas are added to the apneas, and the total is reported as "apnea".

In the export however mixed apneas have their own column. This column does not exits in the 420 export data - they changed the way they presented mixed apnea on screen and simply dropped the mixed column from the export! Talk of sloppiness.

Oh, by the way, the butler did it.
O.

[-SWS]

Don't let go of that bone just yet, Ozi and Glassgal! We still haven't discovered exactly which PB employee was remiss in carrying the mixed apnea count into the exported AI (after having dropped it as a discrete count)!


Neversleeps, that "What the hell are they talking about" thing happens to me when I visit certain foreign countries. Seriously, sorry about all the geek-speak.

[neversleeps]

NO!!! I'm not asking you to apologize for your "geek-speak"!!!!! That would be like asking you to apologize for being smarter than me!! (In which case, the whole world owes me an apology...)

Sometimes I think I'm right there with you, and other times I'm absolutely left in the dust.... But I enjoy trying to keep up!!! I was just feeling a bit embarrassed at my inability to follow and was hoping someone else would chime in and join me in my confusion. But no one did... which is even MORE embarrassing!!! Sigh....

But I'll never give up trying!!!

[rested gal]

Sometimes I think I'm right there with you, and other times I'm absolutely left in the dust.... But I enjoy trying to keep up!!! I was just feeling a bit embarrassed at my inability to follow and was hoping someone else would chime in and join me in my confusion. But no one did... which is even MORE embarrassing!!! Embarassed Sigh....

But I'll never give up trying!!!

LOL, I'm right there with ya, neversleeps! I don't even get embarrassed by my inability to keep up with 'em. I'm used to sitting back watching that kind of brilliance at work. I can't do it myself, but I sure enjoy seeing it happen. I content myself with bringing them dog biscuits and fresh water while they go about doing what they do so well. To paraphrase slightly... "We also serve who stand and wait."

(That was a precious picture, SWS.)

Here ya go, ozi...

[ozij]

Thanks RG! - maybe I'll make that my avatar....(help?)
-SWS -- You mean it wasn't the butler?
NeverSleeps -- I've spent the morning (morning here) on a version that ( I hope) will make things clearer. -SWS is quite right - it's a language problem, not an intelligence problem. And understanding becomes even more complicated when you don't have the software to follow us (I assume you don't yet) and when - as you'll see - PB changes terminology midstream.
Anyway here is my attempt at a detective story for non-techies.

Do you think I should post it on a separate thread as well?

What we are talking about, or: The Case of the Missing Jabberwock

Chapter 1
The background:
Some of us bought our autopaps so they could follow their treatment with the help of the machines' software.
The software reads the raw data collected as we sleep, does calculations, arranges them somehow, and presents the results.
The way data are calculated and the way they are presented on the screen is decided by someone in the company.
That way is not necessarily the best way for doctors or patients who wants to follow their therapy over the longterm.
We might want to look at the data differently. Suppose you own two masks, and have been alternating between them for 4 months. Now you want to compare you AHI with one mask to you AHI with the other. The software will not give you that info.
However if you could get the raw data, you could use excel, add the kind of mask to your data, and the extract the info about how many apneas, leaks, etc you had with each mask over those 4 months. That, and much other info.

Chapter 2
Enter "export":
"Export" is when the software puts out the data into a simple text file that data processing software e.g. excel can read.

Some time in April, one of our posters posted an example of how s/he uses excel to learn much more from her/his data. (I forget who ) I wanted to put my data into it, and learn more about my therapy.

Chapter 3
Discovering the body:
As I was looking at the exported data (i.e. what is supposedly the reliable, raw data) I discovered the the total number of apneas – for the same day and session was different when reported by the export, and when reported on the SL3 screen.
Meaning that one, or both sources were unreliable. There were two different bodies of data, instead of one.

Chapter 4
The inheritance
(In the days of grandad 418P)
There are 3 kinds of apneas: obstructive, central, and mixed. For PB, A
central apnea is a condition where breathing stops, but the machine can identify heart beat oscillations in the mask/hose.

In the days of the 418P, PB used the term "central apnea". They also talked about "mixed apneas" which are a bit of this an a bit of that. Silverlining has a "details" chart which shows a tick for every event (various kinds of apneas, snores, etc) occurring in sleep. They showed tick marks for apneas, hypopneas, central apneas and mixed apneas – using those terms – in the deatails screen.

In the synthesis screen, they obviously decided not to report mixed apnea dicreetely, so in the session view screen, the apnea total is actually the total of apnea+mixed apneas.

In the raw data, the apnea total was for "apneas" alone, and "mixed apneas" had a column of their own.

A bit like Humpty Dumpty: A word means what I want it to mean," total apneas" means one thing on the screen, and another thing in the export.

Chapter 5
The younger generation:New times, new mores
By the time they gave us (well, sold us) the 420e, they made both terminology changes, and software changes.
Terminology:
Suddenly, they no longer have "central apneas" they have "Apneas with CArdiac oscillations" : which in a sense is good – since "central apnea" is a medical diagnostic term, whereas "apneas with cardiac oscillations" is an operational description of what the machine is measuring.

At the same time, they dropped the term "mixed apnea" and no longer have a word for it. For the duration, lets call them "jabberwocks". Jabberwocks are still:
"… a respiratory Apnea consisting of periods when CArdiac oscillations occurred, and periods when there were no CArdiac oscillations).
Software
For the 420E, when a jabberwock happens, it shows on the details screen as an "apnea" and an "apnea c/a" concurrently, there is no discreete "mixed apnea" count, only this composite, unnamed presentation. Jabberwocks are added to the total "apnea" count on the various screens, and subtracted from the "apnea c/a".

But jabberwocks are no longer reported in the raw data. Jabberwocks – aka mixed apneas – have totally disappeared from the raw data.

Chapter 6
Why this should concern non-techies:
Suppose a person had relatively many jabberwocks, sorry, mixed apneas, and suppose the RT, DR. or patient wanted to see the long term effect of therapy on the results: The long term analysis would be skewed. Put simply: The export data is unreliable for any patient with a non-negligeable number of mixed apneas. If most of your apneas are mixed – the export data will be highly misleading – and could cause wrong clinical decisions.

Of course, we don't know how many people have been attempting to use the raw data, what percentage of those had a significant no. of mixed apneas. But we do know that for a long time, SL3 has been sold with a bug.

And now I can finally start thinking of other things...
O.
P.S. For the linguistically and philosphically inclined among us: Don't you think its fascinating to see how the disappearance to the verbal signifier ("mixed apnea") created havoc in the ability to represent the singnified in the data export?...

[-SWS]

-SWS -- You mean it wasn't the butler?

No, I'm with you on the butler part, ozij. This mess surely couldn't have been the work of a computer programmer. I'm just wondering which butler under the employ of PB is responsible. So far it sounds as if the butler who goes by the name Mr. Jabberwocky is responsible...

I agree that this bug could significantly skew a patient's data analysis and resulting treatment decisions. Those who don't export data are exempt from this particular bug and associated risk.

[ahujudybear]

Thanks for the fabulous sleuthing, Ozi! I have been printing out this stuff for my RT and her boss. I'm sure she will very much appreciate the "Non Techie Version!" She just tried to use the program for the very first time the day before she left on vacation. Can't wait to find out just what she has been able to get from the data!

Hope you plan on sending this information to the manufacturer or the programming company?

Hey! They might give you some sort of reward... maybe even hire you?! <LOL>

- JB

[neversleeps]

Cool story, Ozi!! Let me pick up with the next chapter....

Chapter 7
Getting the problem fixed:

The astonishing investigative results brought to light by the MENSA members of the cpaptalk.com forum in the case of the missing jabberwocks was then hand-delivered to Mr. Richard J. Meelia, President and CEO of Tyco Healthcare, by urgent messenger. He chuckled to himself and said under his breath, "People actually look at that stuff?" and then promptly dumped the message in the trash can. Just then a maintenance engineer walked by, and being a compliant and proactive PB420E APAP user, found this matter to be of utmost importance so he quietly stole into the copy room and faxed the information to the folks at the GOSPORT Hampshire UK office, who in turn requested the whole mess be emailed to a Mr. Ales Hrastar, Product Specialist, in Slovenia. (I bet you wish I was making this up....)

Chapter 8
The Denouement

The man in Slovenia, upon receiving the fax exclaimed, "Holy cats!! You mean to tell me an end user discovered this glaring error? Not a physician who was diligently monitoring, compiling data, and conducting statistical analysis for a patient? How very odd...."

Quickly (and I mean, really quickly... like just a matter of minutes... seconds, even...) the glitch was found and immediately fixed.

Chapter 9
The afterward

This historic landmark case was brought to the attention of the Nobel Prize commission. Naturally the resolution of this issue was of tantamount importance, as the treatment of thousands of individuals was compromised due to the heretofore missing jabberwocks.

The committee determined an award was indeed in order! Thus the Nobel Prize was given to, you guessed it, The President of Tyco Healthcare for his invaluable contribution to finding the hidden link.


THE END


The moral to the story: Every jabberwock has a SilverLining.

[ozij]

I'm just wondering which butler under the employ of PB is responsible. So far it sounds as if the butler who goes by the name Mr. Jabberwocky is responsible...

Well, it certainly can't be a butler by the name of Jeeves...

Maybe it was a programmer after all... I ran into this the on Monday, while surfing Mark Minasi's sit for other reasons. I haven't read more than the title page yet.
Mark Minasi on The Software Conspiracy.
Minasi, to those who wonder is a very very professional Microsoft Windows systems pros - has excellent books on the subject -for geeks though - sorry about that.

Sir JB:
I would be very intersted to hear what the program does for the 425... I will have to do translating and explaining for my DME and RT - I'm wondering how to go about it...

NeverSleeps: Great sequel...
O.

_________________

CPAPopedia Keywords Contained In This Post (Click For Definition): DME