Baseline for Desaturations and Flow rates

General Discussion on any topic relating to CPAP and/or Sleep Apnea.
Wondering1
Posts: 37
Joined: Wed Oct 12, 2022 8:02 am

Re: Baseline for Desaturations and Flow rates

Post by Wondering1 » Sun Dec 04, 2022 6:54 pm

When I first glanced at your reply I thought,”well that’s reasonable enough - he is busy with other things” , but then I realized that you think I’ve got too much on MY queue.

My original question (back along time ago) was about baseline definitions. Apparently the baseline computation relies on stability of previous breaths.

You offered a flowrate graphic that I would not consider stable enough to obtain a baseline. You may disagree, that’s OK.
So the obvious question then would be, what is (in your example, in your opinion) the baseline value and how did you derive that value?

So not all that much in MY queue.

User avatar
robysue1
Posts: 897
Joined: Sun Sep 18, 2022 3:39 pm
Location: Buffalo, NY

Re: Baseline for Desaturations and Flow rates

Post by robysue1 » Mon Dec 05, 2022 9:27 am

I went skiing yesterday instead of monitoring this thread that has largely consisted of nonsense posted by Wondering1, with contributions from tech5 and dataq1 along with a quote of some serious nonsense posted over at apneaboard. (Which is a real shame: Most of the time apneaboard doesn't post nonsense .)

But rested and refreshed from the afternoon of skiing and combined with a pretty good night's sleep, I'm back into the mode of "lecture at them 'till they are blue in the face" and "threaten them with This will be on the final exam"
Wondering1 wrote:
Sat Dec 03, 2022 7:20 pm
robysue1 wrote:
Sat Dec 03, 2022 5:15 pm
Actually here's where I have some advantage, I can numerically pick off the peaks for the two minutes prior to the "big breath" and average those.
Likewise I can pick off the peaks for the breaths after the big breath.
There is far more to this than just "eyeballing" the peak numbers.

And without knowing how you eyeballed those numbers, there's no reason to trust your estimates. Give us the screen shots or post the data itself to dropbox so that others can download it and do their own proper analysis.

In other words, the data you actually posted does not appear to agree with your eyeballed estimates. I've actually drawn lines through the max peaks, and the peaks before the big breath are closer to 30 L/min than they are to 24 L/min.

Image

Hence my (conservative) estimate that the running baseline is probably something like 28 L/min. And the smallest peaks in the hypopnea are a bit less than half-way between 12 and 18cm, hence my estimate of 14cm. At any rate, I agree that this is a marginal hypopnea---marginal in the sense that the machine's definition of "hypopnea" has barely, barely, barely been met: The machine has calculated the reduction in flow to be right at the 50% reduction in flow that is needed to trigger being flagged---provided the machine also flagged in something as flow limitation at the same time the flow was reduced. See rubicon's post with images from Resmed that describe exactly when their machines are programed to flag hypopneas.
Wondering1 wrote:
Sat Dec 03, 2022 7:20 pm
The peaks for the breaths following the big breath were 24.90 , 21.58, 15.17, 16.38, 17.75, 19.57, 20.99
The first two breaths after the big breath are not part of the hypopnea; they are part of the baseline "normal" flow. The hypopnea starts where I drew the red line. The breaths towards the end of the hypopnea are "bigger" because the machine is, in fact, looking for a recovery breath or two to make sure the H is actually over.

And again, this hypopnea is marginal: It's right at the edge of the border between "flag this as an H" and "don't flag this as an H". Any numerical algorithm will have difficulties at the margins of where it is applicable. And you've not provided one critical piece of data that the machine is known to use in its decision to flag hypopneas---namely the flow limitation graph.


You have not, and continue to refuse to provide, the all important flow limitation graph, most likely because you are still hung up on the idea that the whole sleep medicine community is somehow wrong to define hypopneas and apneas in terms of the reduction in flow, which is by definition a rate function measured in L/min instead of your pet idea that a reduction in tidal volume (measured in mL or L) ought to be the relevant measure for defining a hypopnea:
Wondering1 wrote:
Sat Dec 03, 2022 7:20 pm
With regard to the TV graph, the two minutes prior to the big breath the TV ranges from 462 to 495 ml.
The moment that the Hypopnea is marked, the TV is 306 ml (between 34 and 38% decrease in the TV
This idea of yours---that TV should be looked at as a criteria for defining hypopneas has been rejected by the entire community of professionals in the world of sleep medicine.

Finally, I'll add: You still haven't complete the homework assignment I gave you way back on page 5 of this thread:
robysue1 wrote:
Fri Dec 02, 2022 11:38 pm
Here's a homework assignment for you:

Read or skim throughThe role of flow limitation as an important diagnostic tool and clinical finding in mild sleep-disordered breathing,
which is a scientific paper about the subtleties and problems of diagnosing sleep disordered breathing.

Now answer these questions for me:

1) How many times is tidal volume mentioned in the entire article? What's the context of when tidal volume is mentioned?

2) Click on the link for Figure 4. What are the names of the three graphs in the figure? And what are the units for each graph in the figure?

3) In light of your answers to Question 2, what units do you think the authors of the paper use to measure what they call "flow" throughout the paper? Do the authors of the paper use the word "flow" in a way that is consistent with thinking of "flow" as a volume function or as a rate function?


And here's a Google assignment for you: Find me a scientific paper on sleep disordered breathing that makes the argument that tidal volume should be used as a criteria for scoring hypopneas on a PSG. Give me the full citation or a web link to the article.
Last edited by robysue1 on Mon Dec 05, 2022 3:35 pm, edited 2 times in total.
Joined as robysue on 9/18/10. Forgot my password & the email I used was on a machine that has long since died & gone to computer heaven.

Correct number of posts is 7250 as robysue + what I have as robysue1

Profile pic: Frozen Niagara Falls

User avatar
robysue1
Posts: 897
Joined: Sun Sep 18, 2022 3:39 pm
Location: Buffalo, NY

Re: Baseline for Desaturations and Flow rates

Post by robysue1 » Mon Dec 05, 2022 9:44 am

dataq1 wrote:
Sun Dec 04, 2022 2:06 pm
Glad to:
The AASM “Chicago consensus paper” states, “Baseline is defined as the mean amplitude of stable breathing and oxygenation in the 2 minutes preceding onset of the event (in individuals who have a stable breathing pattern during sleep) or the mean amplitude of the 3 largest breaths in the 2 minutes preceding onset of the event (in individuals without a stable breathing pattern).” The 2007 scoring manual states, “When baseline breathing amplitude cannot be easily determined (and when underlying breathing variability is large), events can be terminated when either there is a clear and sustained increased in breathing amplitude, or in the case where an oxygen desaturation has occurred, there is event-associated oxygen re-saturation of at least 2%.” The task force recommends that the 2007 manual guideline for determining baseline breathing be upheld
Read the emphasized phrasing.

The AASM understands quite well the problem with establishing baseline amplitude in the flow graph when the patient is experiencing a whole series of sleep disordered events (i.e. hypopneas), one right after the other. But they don't just throw their hands up and say it's no use to use flow to determine hypopneas in this situation.

Rather they have additional proxies that can be used on a PSG to help establish when a hypopnea has occurred, and one of those criteria is the increase in amplitude in the flow graph to mark the end of a particular hypopnea---particularly when the hypopnea does not have a corresponding O2 desat. Tidal volume is not mentioned (ever) as a criteria for scoring hypopneas in PSGs, nor is TV mentioned as a way of determining when a hypopnea is over.

And again, once we restrict ourselves to what xPAPs are (and are not) capable of measuring, we're stuck with flow. And the machine's decision has to be made based on a computer program without a set of human eyes looking over the data and saying, "this thing is too marginal and there's evidence of an arousal, so let's not score it" and also saying "these things barely miss the definition of hypopnea, but there's evidence of unstable breathing and the machine's computed baseline amplitude in flow is suspect specifically because the breathing is not stable; hence these things really should be scored as Hs"

Rather, the scoring/flagging of events by an xPAP's programming is done solely by the numbers, and sometimes the algorithm fails. In other words, sometimes it flags false positives and sometimes it flags false negatives. In terms of efficacy of therapy, the number of false negatives is far more important than the number of false positives: Large numbers of false negatives can imply that therapy is effectively treating the OSA when it is actually not effective and the basic OSA problem remains unresolved.
Joined as robysue on 9/18/10. Forgot my password & the email I used was on a machine that has long since died & gone to computer heaven.

Correct number of posts is 7250 as robysue + what I have as robysue1

Profile pic: Frozen Niagara Falls

User avatar
robysue1
Posts: 897
Joined: Sun Sep 18, 2022 3:39 pm
Location: Buffalo, NY

Re: Baseline for Desaturations and Flow rates

Post by robysue1 » Mon Dec 05, 2022 9:49 am

Rubicon wrote:
Sun Dec 04, 2022 12:28 pm
dataq1 wrote:
Sun Dec 04, 2022 12:03 pm
The evaluation of a hypopnea EVENT is based in a departure from the baseline of STABLE breathing PRIOR to the event.
By-the-by, I missed the part where the somebody said that the baseline must be
STABLE breathing
before the algorithm starts calculating stuff.

Can you point me to that?

TIA.
Yep. That's exactly the point: A numerical algorithm applied by a computer is going to be applied regardless of whether the breathing looks stable or not. And that can cause an xPAP to mis-score events when they're happening right on top of each other. It doesn't always happen, but it can---as the snippets you've posted earlier show.

A knowledgable human being monitoring an in-lab PSG can apply some judgement when the breathing is clearly unstable. And that knowledgable human being also has a whole lotta extra data available to them to back up their decisions to score things as a string of Hs on a PSG even when "baseline amplitude in flow" is difficult, if not impossible, to calculate in a meaningful way.
Joined as robysue on 9/18/10. Forgot my password & the email I used was on a machine that has long since died & gone to computer heaven.

Correct number of posts is 7250 as robysue + what I have as robysue1

Profile pic: Frozen Niagara Falls

User avatar
robysue1
Posts: 897
Joined: Sun Sep 18, 2022 3:39 pm
Location: Buffalo, NY

Re: Baseline for Desaturations and Flow rates

Post by robysue1 » Mon Dec 05, 2022 10:08 am

robysue1 wrote:
Mon Dec 05, 2022 9:27 am
Wondering1 wrote:
Sat Dec 03, 2022 7:20 pm
Actually here's where I have some advantage, I can numerically pick off the peaks for the two minutes prior to the "big breath" and average those.
Likewise I can pick off the peaks for the breaths after the big breath.
There is far more to this than just "eyeballing" the peak numbers.

And without knowing how you eyeballed those numbers, there's no reason to trust your estimates. Give us the screen shots or post the data itself to dropbox so that others can download it and do their own proper analysis.
Alternatively, you could post the same data from Oscar with the following modifications:

1) Change the y-scale on the flow graph so that it is between -36 and +36
2) Increase the vertical size of the flow graph significantly so that it is easier to draw horizontal lines with more refinement
3) Include the flow graph.

In other words, make your plot look similar to this very marginal hypopnea from my data from last night:
Image
Joined as robysue on 9/18/10. Forgot my password & the email I used was on a machine that has long since died & gone to computer heaven.

Correct number of posts is 7250 as robysue + what I have as robysue1

Profile pic: Frozen Niagara Falls

User avatar
robysue1
Posts: 897
Joined: Sun Sep 18, 2022 3:39 pm
Location: Buffalo, NY

Re: Baseline for Desaturations and Flow rates

Post by robysue1 » Mon Dec 05, 2022 11:27 am

Wondering1 wrote:
Sun Dec 04, 2022 6:54 pm
My original question (back along time ago) was about baseline definitions. Apparently the baseline computation relies on stability of previous breaths.
In an ideal world baseline is based on stable breathing.

But the fact is that for folks with significant OSA, the world is far from ideal. On an in-lab PSG, you have a knowledgeable human being looking at multiple channels of data and the tech uses his/her expertise to decide whether anything that is "marginal" should be scored. That tech can also use their knowledge and multiple channels of data to score a series of real hypopneas that are occurring so close together as to make it difficult to formally calculate a baseline based on "stable breathing" and they can easily identify the ends of events based on recovery breaths and O2 resaturation---if the Hs are causing O2 desats in addition to or instead of EEG arousals.

The xPAP machine that sits on the bedside table, unfortunately, does not have a wealth of data available to it. Nor does it have a human being monitoring the data breath-by-breath. What the xPAP has are multiple numerical algorithms for analyzing the flow data that the machine can directly gather. And the programmers who wrote those numerical algorithms have expertise in programming numerical algorithms to analyze flow data on the fly to determine sleep disordered breathing patterns based on what the experts in sleep medicine tell the programmers are important criteria for scoring sleep disordered breathing events.

Both the programmers and the experts involved in designing the xPAP machines know that the algorithms are never going to be "perfect" in the sense that the machine will never score a false positive or a false negative. Rather, the expectation is that if/when xPAP therapy is optimized in an OSA patient, the patient's sleep breathing should be pretty close to normal sleep breathing---i.e. once xPAP is optimized, most of the time the sleep breathing is pretty regular and the machine's algorithm for computing baseline amplitude for the flow graph is accurate enough to keep the number of false negatives (real events that are not flagged) and false positives (non real events that are flagged) down to a statistically insignificant number when you look at the data averaged over a month or more of usage.

That's why the manufacturers of these machines talk about usefulness of trending data instead of obsessing over the daily data. It's also why most sleep doctors care more about the trending data in terms of the AHI rather than one bad night's data with an AHI = 10.something.

But the programs the machines use do require some kind of a baseline to be computed from the available data, and while the programmers can take care of some standard scenarios of unstable breathing in computing a baseline, they can't account for every possibility that can lead to unstable breathing and correct the baseline computation appropriately. And so the person's breathing is not particularly stable, screwy things can happen in terms of "missing" events. We mostly see this when a person's OSA is not effectively controlled by their xPAP or when a person has a significant problem with CSA that manifests itself with periodic breathing associated with hyperventilation/hypoventilation cycles. Sometimes when a whole string of sleep disordered events occur in a very short time, the machine's calculated baseline for flow amplitude can incorrectly cause the calculated baseline to be too low and that can result in the machine not flagging a whole string of events that ought to be flagged. Sometimes with a whole string of central hypopneas there won't be a characteristic flow limitation and the machine won't score the hypopneas. (And in this case, it's a good thing that the machine is not scoring the hypopneas, since that prevents the machine from unnecessarily increasing the pressure, which could/would make the string of central stuff get worse.)

These algorithmic shortcomings do not represent a problem with the basic definitions of apnea and hypopnea which are in terms of a sudden reduction in flow: They're simply a result of the xPAP being forced to use an algorithm in a setting where the algorithm's built-in limits are being hit with what's happening in real time with respect to the PAPer's now unstable breathing.
Joined as robysue on 9/18/10. Forgot my password & the email I used was on a machine that has long since died & gone to computer heaven.

Correct number of posts is 7250 as robysue + what I have as robysue1

Profile pic: Frozen Niagara Falls

Wondering1
Posts: 37
Joined: Wed Oct 12, 2022 8:02 am

Re: Baseline for Desaturations and Flow rates

Post by Wondering1 » Mon Dec 05, 2022 8:28 pm

robysue1 wrote:
Mon Dec 05, 2022 9:27 am
I went skiing yesterday instead of monitoring this thread that has largely consisted of nonsense posted by Wondering1, with contributions from tech5 and dataq1 along with a quote of some serious nonsense posted over at apneaboard. (Which is a real shame: Most of the time apneaboard doesn't post nonsense .)

But rested and refreshed from the afternoon of skiing and combined with a pretty good night's sleep, I'm back into the mode of "lecture at them 'till they are blue in the face" and "threaten them with This will be on the final exam"
Good that you got some rest and relaxation.
Here's how I acquired the peak flow values for both the baseline and after the big breath:
I read then off the OSCAR graph for each inhalation (30 of them) prior to the big breath:
Image
I repeated the same process for each breath after the big breath (seven of them) and then compared each of the seven breaths to the baseline average.
Those values and their respective change in % was what I quoted above.

Short of accessing the EDF files and searching for the pertinent breaths, I can think of anything else to do. So I hope that you can find that trust worthy enough.

Now perhaps I've misunderstood you, so allow me to feedback to you what I think you have been saying all along.

When assessing inhalation curves for performance the PEAK AMPLITUDE (of the flowrate curve) is the critical element, regardless of the duration of that peak.

At least that's what I've been understanding of your position.

My position has been that the area "under the curve" for an inhalation is at least equally important, if not more so.

Dataq's schematic, @ viewtopic/t185583/Baseline-for-Desatura ... 0#p1428303, crude though it might be illustrates this.

I would anticipate that the TV would represent the "area under the curve", thus the flow of air into the lungs with each inhalation.

Wondering1
Posts: 37
Joined: Wed Oct 12, 2022 8:02 am

Re: Baseline for Desaturations and Flow rates

Post by Wondering1 » Mon Dec 05, 2022 8:54 pm

robysue1 wrote:
Mon Dec 05, 2022 11:27 am
I frankly haven't read all of your recent post-sking posts yet,
but I did want to remark, although it has no bearing on the peak-amplitude versus volume discussion, the Flow Limitation aspect was interesting enough that I went to look if there was some significant flow limitation occuring during the dubious hypopnea. The flow limitation was 0.04, so not so much of a factor there.

User avatar
Rubicon
Posts: 1678
Joined: Sun Feb 20, 2022 6:59 am

Re: Baseline for Desaturations and Flow rates

Post by Rubicon » Tue Dec 06, 2022 12:06 am

robysue1 wrote:
Mon Dec 05, 2022 9:27 am
I went skiing yesterday ...
I'm glad you're able to do something with that 9 feet of snow!

I too am taking a break from-- well, taking a break. Yes Pugsy, that's a real IP, not a VPN! Soon going a little more to the east, where it'll be a bucket-list place to spend Christmas (clue: not in o.'s neighborhood, although that obviously would be the first place one would guess)(hmmm. Perhaps a thought for next year)
Freeze this moment a little bit longer.
Make each sensation a little bit stronger.
Experience slips away.

dataq1
Posts: 816
Joined: Sat Dec 24, 2005 8:09 am
Location: Northeast Ohio

Re: Baseline for Desaturations and Flow rates

Post by dataq1 » Tue Dec 06, 2022 12:40 pm

Wondering1 wrote:
Mon Dec 05, 2022 8:28 pm

Here's how I acquired the peak flow values for both the baseline and after the big breath:
I read then off the OSCAR graph for each inhalation (30 of them) prior to the big breath:
Actually Wondering made a slight error in his process. He should have included the value from the "big breath"
The evaluation process used by the machine doesn't know to exclude an errant (non-typical) breath when coming up with an average.

So, include the "big breath" peak in computing the average of the previous 2 minutes, before comparing to each subsequent breath.
"THE INFORMATION PROVIDED ON CPAPTALK.COM IS NOT INTENDED NOR RECOMMENDED AS A SUBSTITUTE FOR PROFESSIONAL MEDICAL ADVICE."

Wondering1
Posts: 37
Joined: Wed Oct 12, 2022 8:02 am

Re: Baseline for Desaturations and Flow rates

Post by Wondering1 » Tue Dec 06, 2022 4:11 pm

dataq1 wrote:
Tue Dec 06, 2022 12:40 pm

So, include the "big breath" peak in computing the average of the previous 2 minutes, before comparing to each subsequent breath.
Will do later tonight

User avatar
Rubicon
Posts: 1678
Joined: Sun Feb 20, 2022 6:59 am

Re: Baseline for Desaturations and Flow rates

Post by Rubicon » Wed Dec 07, 2022 12:23 am

Seems to be bogging down a little.

The discussion alternates between using AASM and ResMed rules. As it appears that your event is post-arousal in nature, and with no oximetry to view, under AASM rules it would not be scored as anything.
Rubicon wrote:
Thu Dec 01, 2022 3:14 pm
If we review the patent it's all about the RMS of flow (your department) with only a peripheral mention of tidal volume "in some embodiments" (but I think we can agree it's not this one).
The tidal volume graph that you're looking at is based on a 5-breath moving average. You might want to consider that in doing breath by breath calculations.
Freeze this moment a little bit longer.
Make each sensation a little bit stronger.
Experience slips away.

Wondering1
Posts: 37
Joined: Wed Oct 12, 2022 8:02 am

Re: Baseline for Desaturations and Flow rates

Post by Wondering1 » Wed Dec 07, 2022 11:06 am

dataq1 wrote:
Tue Dec 06, 2022 12:40 pm
So, include the "big breath" peak in computing the average of the previous 2 minutes, before comparing to each subsequent breath.
Ok, so did that (same process in acquiring (reading) the data and the baseline value should be (with the big breath included) which altered the baseline slightly from 24.68 (original calc) to 24.71.

However, Robysue suggested (and I must say, rightly so) that the baseline is not static. That is to say, when the next breath is being evaluated, the baseline is the 30 breaths prior to the breath under evaluation. (so the baseline value changes with successive breath).

Using that process (a non-static baseline) the peak amplitude for each of the breaths changed in percent by a small amount, for example what I had previously calculated as 38.53 should be 38.42.

After doing all the recalculations what I had originally said still stands, only two of the breaths following the big breath only two crossed the 30% threshold, one at 38% and one at 33%.

User avatar
robysue1
Posts: 897
Joined: Sun Sep 18, 2022 3:39 pm
Location: Buffalo, NY

Re: Baseline for Desaturations and Flow rates

Post by robysue1 » Wed Dec 07, 2022 11:28 am

Rubicon,

A big thank you for your comments!
Rubicon wrote:
Wed Dec 07, 2022 12:23 am
Seems to be bogging down a little.

The discussion alternates between using AASM and ResMed rules. As it appears that your event is post-arousal in nature, and with no oximetry to view, under AASM rules it would not be scored as anything.
Thanks! My initial response to Wondering1 way back on page 2 included this:
robysue1 wrote:
Thu Dec 01, 2022 2:10 pm
Should it have been flagged? It actually looks like 2, maybe three, sub optimal breaths, taken after a deep breath.
To my nonprofessional's eyes, this could be a post arousal thing (i.e. SWJ) that might not have been scored by a tech looking at all the channels of data available during a PSG.

You also write:
Rubicon wrote:
Wed Dec 07, 2022 12:23 am
The tidal volume graph that you're looking at is based on a 5-breath moving average. You might want to consider that in doing breath by breath calculations.
Thanks for that tidbit of information that is highly relevant to this thread's main point of contention.

It had become clear to me that what was graphed in the tidal volume graph was some kind of a moving average, but I had no idea of what the framing of the moving average was. A 5-breath moving average of TV makes it clear exactly why the "big bump" happens after the big inhalation happens.

It also calls into question the whole idea that Wondering1 and dataq1 are postulating: That somehow the tidal volume graph should be used to flag apneas. Clearly it should not.
Joined as robysue on 9/18/10. Forgot my password & the email I used was on a machine that has long since died & gone to computer heaven.

Correct number of posts is 7250 as robysue + what I have as robysue1

Profile pic: Frozen Niagara Falls

User avatar
robysue1
Posts: 897
Joined: Sun Sep 18, 2022 3:39 pm
Location: Buffalo, NY

Re: Baseline for Desaturations and Flow rates

Post by robysue1 » Wed Dec 07, 2022 11:29 am

Rubicon wrote:
Tue Dec 06, 2022 12:06 am
robysue1 wrote:
Mon Dec 05, 2022 9:27 am
I went skiing yesterday ...
I'm glad you're able to do something with that 9 feet of snow!
We're in the northern half of the metro area. We got a modest 12-15 inches of snow and it's long since gone.
Joined as robysue on 9/18/10. Forgot my password & the email I used was on a machine that has long since died & gone to computer heaven.

Correct number of posts is 7250 as robysue + what I have as robysue1

Profile pic: Frozen Niagara Falls