Baseline for Desaturations and Flow rates

[Wondering1]

But my main question is this..... when examining these charts for hypopneas, should the tidal volume chart be used for guidance?

I think you should blow up the chart if you want to measure what seems to ne a marginal hypopnea. Single channel, 60 second window, fit to window like:

Robysue kindly quoted from the Resmed Clinical manual:

A hypopnea is flagged "when the respiratory flow decreases to 50% for at least 10 sec."

What I noted there was that it said respiratory flow, but it did not say respiratory flow rate
That's a clue that that it's the volume of air, not the flowrate, that is determinative.

That's what leads me to speculate that it's the tidal volume graph (representing the volume) that needs to be examined.



.

[Rubicon]

Robysue kindly quoted from the Resmed Clinical manual:

A hypopnea is flagged "when the respiratory flow decreases to 50% for at least 10 sec."

What I noted there was that it said respiratory flow, but it did not say respiratory flow rate
That's a clue that that it's the volume of air, not the flowrate, that is determinative.

That's what leads me to speculate that it's the tidal volume graph (representing the volume) that needs to be examined.

So here's what I need you to do:

Hit yourself upside the head with this 3 times:



Reread what you just wrote.

Then come back and tell me if it still makes sense.

[Wondering1]

Reread what you just wrote.
Then come back and tell me if it still makes sense.

So I reread, and it still makes sense.
The distinction is that the Flowrate graphs (from OSCAR) are the RATE of flow.
The Tidal Volume graphs (from OSCAR) are the VOLUME of flow.

The Resmed definition (provided by Robysue) refers to flow quantity not the rate of flow.

Can you clarify, what is the issue as you see it.

[Rubicon]

...what is the issue as you see it.

You're probably not swinging the hammer hard enough.

[Wondering1]

...what is the issue as you see it.

You're probably not swinging the hammer hard enough.

Could you try to be serious?
Exactly what is your issue with what I've said?
I don't understand what your trying to say.

[robysue1]

Above I posted a flowrate curve with a flagged hypopnea, in considering what Robsue1 posted, is it best to examine the tidal volume / time graph to understand why a hypopnea was flagged?

I haven't the foggiest idea. But I'm inclined to say, "No, there's no need to examine the tidal volume graph to understand why a hypopnea was scored." The tidal volume data is calculated from the flow data by the xPAP as near as I can tell. And it's certainly not "sampled" at the same level of frequency, so it's less high resolution data than the flow data is.

When I see things things that look like this:



I tend to regard them as cases at the edges of the algorithm ability to detect hypopneas---i.e. in any kind of numerical algorithm, things that barely meet or barely miss meeting a particular for defining a "flag" are going to have some instances that look questionable because they questionable. If you stare at enough flow graphs, you see "events" that you wonder why they were scored (like this one) and you will also see places in the flow graph where you wonder why something was not flagged by the machine as an apnea or a hypopnea.

My guess is that the particular stretch of breathing in your example barely, barely, barely meets your xPAP machine's definition of an H, possibly because that big, big breath that ends at 5:55:40 could be just big enough to screw up the computation of the baseline flow rate amplitude. And there are 3-5 individual breaths that make up the H that are clearly quite a bit smaller than the stable breathing both before the big breath ending at 5:55:40 and the stable breathing after the H is over. Whether the amplitude of those small breathes is genuinely represents a 50% decrease (Resmed definition of an H) from a possibly slightly elevated "baseline" is almost impossible to tell---not without drawing a bunch of horizontal lines and then zooming in even more on those questionable breaths. And by zooming in, I mean zooming in on both the x and y axes: The current y-scale makes it difficult to see what the amplitude of the inhalations actually are during the breaths that are flagged as an H.

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. Or it could be something that a tech would score because all those missing channels of data indicate you are actually asleep when this happened and the decrease in flow rate meets the AASM standard for scoring a hypopnea. There's just no way to tell with 100% accuracy whether this could/should/would have been scored on an in-lab PSG.

Or perhaps you are thinking this event couldn't have lasted for 10 seconds. But 2 or 3 sub optimal breaths can easily meet the "10 second" rule---if I recall correctly, the 10 seconds start with the nadir of the last (normal) exhalation and end with a breath that is big enough to be close to "baseline" or bigger than baseline.

At any rate, xPAP machines only have one channel of direct data available to them: The flow of air into and out of your lungs. And the programmers have made decisions about what to flag in ways that are designed to catch as many "real" events as possible without catching too many "non-real" events. That's why no manufacturer claims that the data is 100% accurate. Rather, it's accurate enough for trending data to be valid enough to determine efficacy of xPAP therapy: If your machine-scored AHI is (well) below 5 over the course of a month or more, that's solid evidence that not too many obstructive events are occurring on a regular basis.

[Rubicon]

Could you try to be serious?

No.

Whether the amplitude of those small breathes is genuinely represents a 50% decrease (Resmed definition of an H) from a possibly slightly elevated "baseline" is almost impossible to tell---not without drawing a bunch of horizontal lines and then zooming in even more on those questionable breaths. And by zooming in, I mean zooming in on both the x and y axes: The current y-scale makes it difficult to see what the amplitude of the inhalations actually are during the breaths that are flagged as an H.

Well I showed said person what was needed. Request ignored.

As well as my digressive image, which IMO is a critical part of this discussion.

I haven't the foggiest idea. But I'm inclined to say, "No, there's no need to examine the tidal volume graph to understand why a hypopnea was scored." The tidal volume data is calculated from the flow data by the xPAP as near as I can tell. And it's certainly not "sampled" at the same level of frequency, so it's less high resolution data than the flow data is.

Agree.

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).

[robysue1]

Whether the amplitude of those small breathes is genuinely represents a 50% decrease (Resmed definition of an H) from a possibly slightly elevated "baseline" is almost impossible to tell---not without drawing a bunch of horizontal lines and then zooming in even more on those questionable breaths. And by zooming in, I mean zooming in on both the x and y axes: The current y-scale makes it difficult to see what the amplitude of the inhalations actually are during the breaths that are flagged as an H.

Well I showed said person what was needed. Request ignored.

I never meant for you to draw all those horizontal lines. However, some of them are in fact drawn in the image you included.

And I guess I'm just a sucker for giving folks long winded discussions with more detail than they asked for because of the 40 years I've spent teaching freshmen math students Calculus. The only way the Calculus sticks for some of them is for me to explain it over and over and over and over and over ....

And to break it down into little bitty bits.

In other words, I tend to prefer hitting people over the heads with a two-by-four instead of giving them a hammer to hit themselves with.

As well as my digressive image, which IMO is a critical part of this discussion.

Agreed. If I'd had the time, I would have included that in my post since your image is exactly what I meant by drawing lots of lines and zooming in on both the x- and y- directions.

I haven't the foggiest idea. But I'm inclined to say, "No, there's no need to examine the tidal volume graph to understand why a hypopnea was scored." The tidal volume data is calculated from the flow data by the xPAP as near as I can tell. And it's certainly not "sampled" at the same level of frequency, so it's less high resolution data than the flow data is.

Agree.

Glad to see we're on the same page here.

I really think wondering1's main problem is that he doesn't understand that "flow" is a "rate" function. And that calling the flow graph the "flow rate" graph is redundant. And that the hypopnea definitions (AASM as well as Resmed and PR) are all based on a reduction in the flow.

[Wondering1]

While I won't disagree with much of what you've posted, perhaps we misunderstanding each other when talking about flows and flowrates.

I'm referring to flow rates as shown in the OSCAR flow rate graphic and whos units are l/min.
When I refer to flow I'm referring to the tidal volume graphic with the units ml.

Of course they are related, however to get to volume from flow rate the area (from each breath flow rate) has to be calculated. (which explains why the data rate for tidal volume is "slower' than that of the data rate for flow (if the only sensor available is flowrate)

viz:
in spite of the flow rate on each of these breaths is nominally equal (actual ~ 27 L/min), the volume of these breaths are significantly different by about 25%.

The Tidal volume chart in OSCAR displays that difference, while simply the height of the flowrate curve shows no practical difference.

That is why I'm suggesting that attention needs to be paid to the tidal volume when looking at changes in flow (volume).

[Rubicon]

...to get to volume from flow rate the area (from each breath flow rate) has to be calculated. (which explains why the data rate for tidal volume is "slower' than that of the data rate for flow (if the only sensor available is flowrate)

That's not why.

The calculation could be instantaneous if they wanted it to be. AAMOF, in many applications (like volume ventilation) it is.

[Rubicon]

Glad to see we're on the same page here.

What, we're always on the same page! We are like the same page!

Check out all these obvious hypopneas that... aren't:

[robysue1]

Reread what you just wrote.
Then come back and tell me if it still makes sense.

So I reread, and it still makes sense.
The distinction is that the Flowrate graphs (from OSCAR) are the RATE of flow.
The Tidal Volume graphs (from OSCAR) are the VOLUME of flow.

The Resmed definition (provided by Robysue) refers to flow quantity not the rate of flow.

Can you clarify, what is the issue as you see it.

By definition flow is a rate function, not a quantity function. The flow is the rate (in L/min) that air is moving into or out of your lungs at time t. In math-speak, flow is the derivative of the the volume of air in your lungs at time t. When you are inhaling, the volume of air in your lungs is increasing (so the flow is positive) and when you are exhaling, the volume of air in your lungs is decreasing (so the flow is negative).

Saying "rate of flow" is like saying "rate of speed". It's redundant, but sometimes the redundant phrase creeps into the language even though it's not needed.

In ResScan (the official ResMed software), the graph that Oscar labels as "Flow Rate" is labeled as "Flow." In Encore Pro (the official PR software), the same graph is called the "Wave Form" graph.

And the AASM definition of hypopnea, as well as the Resmed definition of hypopnea and the PR definition of hypopnea all are based on the data that is displayed in the graph that Oscar somewhat redundantly labels the "Flow Rate" graph.

And the tidal volume graph, which you seem to think is needed to figure out hypopneas, is a lower resolution graph than the Flow graph is AND the tidal volume numbers are calculated by using the flow data.

Hence, to figure out why something is flagged as an H, there's no need to look at the lower resolution tidal volume graph using calculated numbers based on the data from the flow graph itself.

The problem, such as it is, is that Oscar (and SleepyHead before) (and SleepHQ as well) misnames the graph with the redundant phrase "Flow Rate" instead of the simpler, yet just as accurate, "Flow".

[robysue1]

While I won't disagree with much of what you've posted, perhaps we misunderstanding each other when talking about flows and flowrates.

I'm referring to flow rates as shown in the OSCAR flow rate graphic and whos units are l/min.
When I refer to flow I'm referring to the tidal volume graphic with the units ml.

Of course they are related, however to get to volume from flow rate the area (from each breath flow rate) has to be calculated. (which explains why the data rate for tidal volume is "slower' than that of the data rate for flow (if the only sensor available is flowrate)

viz:

Ah, the chance to talk about the definite integral in a public forum full of non mathematicians. Can I keep my heart from going pitter-patter?

First, however, we have to be clear about vocabulary: The graph that Oscar calls a "Flow Rate" graph is a graph of the flow function, and the flow function itself is a rate function, not a quantity function. (ResScan correctly labels this graph as the flow graph. In other words, the "flow function" and the "flow rate function" are one and the same thing. They are both measured in L/min and they are both a measure of the rate air is moving into and out of your lungs.

And the AASM definition of hypopnea, along with the Resmed definition for flagging a hypopnea and the PR definition, are defined in terms of a reduction in the flow graph---i.e. they're all defined in terms of the heights of the inhalations being significantly reduced from a running baseline presumably calculated on about 2 minutes of prior breathing.

Yes, the red shaded areas in the figure you included do represent the volume of air that you inhale on on each breath. Mathematically they are found by integrating the flow function over the time period of the individual inhalation. But the flow function is, in reality, a curve drawn by connecting a whole lotta data points, rather than being defined by an algebraic function. So finding the definite integrals to calculate the volume of air inhaled for each and every inhalation would require sampling the data (at a lower resolution) and using something like Simpson's Rule to approximate the definite integral. All of this introduces additional various types of numerical errors (with the biggest one being plain old roundoff error. And for what purpose?

If the flow data is not accurate enough to reliably detect a hypopnea, then calculated data based on the flow data is even less reliable for detecting a hypopnea because the calculated data will have more errors in it than the original data does.

The Tidal volume chart in OSCAR displays that difference, while simply the height of the flowrate curve shows no practical difference.

As both Rubicon and I have indicated, the heights in the flow graph DO show a very practical difference: They are significantly shorter than the stable breathing before the big inhalation that ends at 5:55:40 as well as the stable breathing that follows the end of the H.

And if I had access to that flow data in Oscar or ResScan, I could zoom in on those breaths, change the y-axis scale, take a screen shot and plot some appropriate horizontal lines that would very likely show that the amplitudes of the 2-4 worst breaths do, in fact, display about the 50% decrease in amplitude needed to satisfy the definition of hypopnea used by a Resmed machine.

That is why I'm suggesting that attention needs to be paid to the tidal volume when looking at changes in flow (volume).

You may be suggesting it, but the world of sleep professionals seems to disagree with you.

[robysue1]

Glad to see we're on the same page here.

What, we're always on the same page! We are like the same page!

Check out all these obvious hypopneas that... aren't:

Serious question: Any speculation on why those "obvious hypopneas that ... aren't" are not flagged?

Is it because the machine's interpretation of "baseline" flow is being inappropriately depressed by including the multiple hypopneas that aren't in the computation of "baseline"? And if the "baseline" is too low, then those hypopnea breaths might still be just a wee, tiny bit above the bad "baseline" number?

I assume that a tech scoring this kind of stuff on an in-lab PSG with lots more data would also be taking into account the difficulty of establishing a meaningful baseline in terms of "the last two minutes" of breathing here. Of course an xPAP doesn't have a little human being in it who can say, "This looks fishy; maybe I should score these since the baseline might be misleading?"

[Rubicon]

Serious question: Any speculation on why those "obvious hypopneas that ... aren't" are not flagged?

Yup. Because ResMed won't score hypopneas in the absence of flow limitation (at least one).

I assume that a tech scoring this kind of stuff on an in-lab PSG with lots more data would also be taking into account the difficulty of establishing a meaningful baseline in terms of "the last two minutes" of breathing here.

Also, with effort belts, they could also differentiate obstructive from central hypopneas.

Anyway, the OSA Sleep World doesn't care about tidal volume cause they don't treat volume-- they treat restriction to flow.

I suppose the CSA/CompSAS Sleep World treats volume, but they look at flow too.