HYPOPNEA verification

[dataq1]

@ Dave,
Not germane to the original post.
We've never met with this doctor before (prior doc moved out-of-country), and in setting up our initial appointment  we told his office that we were experimenting with a new mask system. Their request was to keep everything the same for two weeks before our first visit, so we could look at a stable system of settings. Not an reasonable request.  

I don't know how you get the notion 

they don't trust your setting judgement

All they seem to be asking is: Please refrain from spinning the dials until we see your wife.

When I said said "his" experiment, it was really "his" experiment to remain in a stable condition.

[dataq1]

Very interesting interaction with this new doctor. He's a youngish (38 yo) MD whose been in the sleep field for 13 years and there are 3 others in the practice. 
He was quite interested in the 30 days of OSCAR reports that I brought, and bemoaned that the Airview/ResScan data that is available to him (from Resmed) is not as detailed as that of OSCAR - when he wants to take a closer look.

I was somewhat surprised when he compared Resmed Flow Lmitations grading system to Voo-Doo science !  

He was insufficiently impressed with Resmed's assessment of Cheyne-Stokes events, (aka periodic breathing), and noted that it was (for my wife) paroxysmal with a 30 burden of less than 3%. Coupled with a recent echocardiogram that was unremarkable, and no other signs of CHF, he felt it was safe to do nothing and revisit in another 90 days, whereupon he asked us just to drop off a duplicate of her then current SD card. 

He also offered that among his local sleep doctors (we are in a  mid-sized city) that their working group had tried "for years" to get Resmed engineers to clarify exactly the methodology they apply to classify hypopneas.  After years of vague  or fuzzy discussions, some of the working group have suggested that 1) they developed an algorithm some years ago and that programmer did not adequately document what his coding actually did and has subsequently either died or left the company, or 2) they just do not want anyone to know that they can't explain it.

WOW!

So, I asked the question: If you consider the data ( for example, AHI) unreliable, why do you guys rely on it?
His answer was simple: Because, short of a PSG, it's the only tool we have.

He went on to say that if consumer-grade XPAPS incorporated oximetry, that would be a substantial step forward towards confidence. (provided that the oximetry system were reliable)

[Crimson Nape]

In case you are unaware, the AS11 now accepts data from the Nonin 3150 BLE Wrist-Worn Pulse Oximeter. Resmed push out an over-the-air update a few months back. Of course, it's $970 USD.

- Red

[Sleeprider]

The easy answer is that Resmed scores hypopnea as follows: Hypopnea is a partial blockage of the airway. During a hypopnea, breathing is shallow at 30%-50% less than your normal breathing for 10 seconds or longer. Above 50% reduction an apnea is scored, and any reduction below 30% is ignored. https://ap.resmed.com/knowledge/hypopnea

[dataq1]

The complications begin with what constitutes "a breath", is it the volume of air or is it the rate?

Once having clarified that, how many breaths satisfy the 10 second criteria?
Assuming that an average patient is breathing at 15/min, 10 sec would be a measurement of 2.5 breaths. So does the software only examine 2 consecutive breaths or 3 consecutive breaths. 

Once you decide to examine (and measure) the "average" of the 2 or 3 breaths, next it would seem that you want to compare that "average" with the corresponding average of breaths just prior to the shallow breath segment. 

So how many breaths (or seconds) do you need to examine to get an "average" prior to the shallow breath segment.

Once having established a minimum 30% reduction does the software mark the end of the shallow section or is there a delay where the "mark or flag" is not at the end of the shallow segment but rather some moments/seconds later?

Unfortunately the AASM standard criteria also fails to address these questions. 

But those questions had to have been addressed in order to code the algorithm.

BTW, you have my word as a gentleman that I'm not trying to be argumentative, I'm only trying to get a clear answer.
Someone, somewhere in the Resmed universe knows the answers; alternatively someone once knew and now has forgotten.

[Sleeprider]

I'm not seeing this as argumentative, but the practical application of an algorithm to the real-time decisions on how to treat someone seems to be of limited usefulness. It is argumentative because you have associated it with the therapy of your friend, so your first post seemed to seek a solution, and now it's mostly to learn Resmed's algorithm. It's good to be acquainted with the specific mathematical basis of the algorithm for flagging hypopnea, but it is more practical to accept that it is what it is and act on that using techniques known to mitigate the problem. I have not found the answer to your question specifically, but I'll instead offer a hypothesis that should be able to be tested. The ASV algorithm targets a minute vents that is within 90% of the 3-minute moving average. It seems reasonable to assume that hypopnea is similarly based on a moving average of either minute vent or tidal volume falling between a 30% to 50% reduction.

Resmed measures hypopnea based on what it can detect with the flow sensor. This is not equal to what AASM defines as hypopnea which is associated with 3% or 4% desatuation of SpO2. Event that metric is not further interpreted in the clinical setting to identify whether it central or obstructive. A Hypopnea is a minimum SpO2 oxygen desaturation according to AASM and a percent reduction of breathing volume for at least 10 seconds according to Resmed...apples and oranges. Treat the hypopnea, not the method.

[BoxcarPete]

To make things more fun, take a look at this lovely cluster from my chart last night. My airsense 10 can't tell the difference between RERA and Hypopnea under thsese circumstances, and there's even an unscored hypopnea in the middle there. My mean TV for the night was 480, and from 4:19:20 or so there are 17 seconds where OSCAR is reporting calculated TV below 337, which is 70% of the nightly median followed by 7 seconds at or above 480. So presumably the preceding cluster starting just prior to 4am influenced the calculation to some extent.

And if you want to get AASM involved, based on previous experience I expect that if I were wearing an oximiter it would not have moved more than 2% at any point throughout that whole cluster.

[dataq1]

Thanks Sleeprider for your thoughts on this, I do appreciate the time! 

If I read you correctly you are saying that (based on the ASV application) that the compared segments are probably volume based rather than peak amplitude based.

Meaning that the flow rate plot needs be evaluated such that the area under the curve of each breath needs to be calculated. (not that difficult to do). 
That would make sense to me, as the concern is more likely how much (aka volume) air is being passed as versus the instantaneous rate of air being passed for each breath (or groups of breaths). 

Are you in (at least speculative) agreement with that?  


BTW, is your handle only coincidentially suggestive of "Palerider" on another channel?

[Sleeprider]

Flow rate leads you to volume in time. I do think the a reduction of 50% to 70% of tidal flow volume over at least 10-seconds (same as 30% to 50% of baseline) results in the flag rather than peak flow, but either could be measured. Might be a good question to ask the developers that actually look at the Resmed data for Oscar.

There is no resemblance between myself and Palerider, who is also on this forum, but mostly in the Oscar development area. I am also in the CPAP talk if you go to look. He would not be replying to you civilly.

[dataq1]

He would not be replying to you civilly.

Have a flare for understatement do you ! 

Meanwhile I'm diving into the Tidal Volume data from the EDF files. (have a sneaking suspicion that resmed's TV may be calculated over more than one breath - we'll see)