Apnea Board Forum - CPAP | Sleep Apnea
Why don't these flow rate curves count as flow-limited? - Printable Version

+- Apnea Board Forum - CPAP | Sleep Apnea (http://www.apneaboard.com/forums)
+-- Forum: Public Area (http://www.apneaboard.com/forums/Forum-Public-Area)
+--- Forum: Main Apnea Board Forum (http://www.apneaboard.com/forums/Forum-Main-Apnea-Board-Forum)
+--- Thread: Why don't these flow rate curves count as flow-limited? (/Thread-Why-don-t-these-flow-rate-curves-count-as-flow-limited)

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13

Why don't these flow rate curves count as flow-limited? - cathyf - 06-25-2021

This is one of those {I thought that I understood this stuff but now I'm saying "huh?" moments!}

Here is a short bit from last night:

The tops of all of those inspiration curves are all jaggy, but the flow limit says zero except for those 4 little bumps which are only 0.04, 0.03, and the two on the right both 0.01 -- in other words the machine is not calling any of that flow limits. I mean, really, if you want to talk about M-shaped waves when you look at those curves you can almost HEAR the "ommmmmmmmmmmmmmm" LOL

Do you think that the machine is taking those bumps during the pauses on the expiratory phases and looking at those shapes too when it's figuring flow limits? There are obviously a lot more of those and I could see that it would swamp the signal from the real inspiratory bumps.

(For some reason when I see those curves with the drawn-out-jaggy-just-below-zero ends of the expirations I see llamas -- or maybe alpacas? Especially with the M shapes on the inspiratory waves which look like the ears sticking up -- I think we should call this waveform "llama breaths" !)

RE: Why don't these flow rate curves count as flow-limited? - Geer1 - 06-25-2021

These are basic machines with basic programming. Those are not simple obvious flow limitations so they don't get reported as such.

Not sure that those are flow limited breaths, almost looks like large amplitude cardiogenic oscillations or something being imposed onto flow rate. Usually only see cardiogenic oscillations in the pause between exhalation/inhalation but looks like something is imposing little air flow spikes even during inhalation. Cardiogenic oscillations are usually caused by blood flow based on my research. Could also be some sort of slow frequency snore like action. Maybe muscle contractions of some sort.

Maybe some sort of restricted breaths but doesn't necessarily look that way imo.

RE: Why don't these flow rate curves count as flow-limited? - Geer1 - 06-25-2021

These are obvious flow limited breaths. You can see a couple very small cardiogenic oscillations prior to inhalation on a couple breaths as well. Different waveform then yours.


RE: Why don't these flow rate curves count as flow-limited? - cathyf - 06-25-2021

Editting this to add another picture...

Here's 3 48-second closeups, two without flow limits, one with -- I'm not understanding what's different...


I'm not seeing them as being different in character? What am I missing?

RE: Why don't these flow rate curves count as flow-limited? - Geer1 - 06-25-2021

What you are missing is that these are basic machines with basic programming incapable of scoring flow limited breaths to the level that you believe they are. Your closer up views do make me think there is some sort of potential flow limitation going on but I don't get that by analyzing the single breath waveform like these machines likely do, I get that by looking at the group of breaths as a whole and thinking they have a flat topped nature to them. A flat topped nature that seems to be interrupted by some sort of other flow blip not only at beginning and end of breath as is most common in flow limited breaths but also sometimes randomly in the middle of breaths. These machines do not have the artificial intelligence level ability required to draw conclusions like this.

Look at my example again (which imo is far more representative of flow limited breathing) and you will see that it didn't score flow limitations from 3:08:30 to 3:08:50. Flow limitations are an imperfect measurement that you have to take with a grain of salt and interpret on your own. The values on average can help understand if breathing is flow limited but it isn't anywhere near accurate to analyze on a breath by breath basis like you are trying to do.

RE: Why don't these flow rate curves count as flow-limited? - 2SleepBetta - 06-28-2021

cathyf wrote:

Here's 3 48-second closeups, two without flow limits, one with -- I'm not understanding what's different...

[Cathyf's graphic was posted by her, here. It was since removed. See it below, after my, 2SB's, edit making it available for understanding comments about it in post below]

I'm not seeing them as being different in character? What am I missing?

...Just before posting this I saw another of your probing questions in reply to my earlier and shortest ever post, today, in Software's the "flow limit expression" thread. Gotta read it more closely, but think this touches on the matters as in the quash down (decay of a FR) post of yours and, possibly, it may do so regarding your FL and their persistence. More at some point later, but need to get an update into my thread re FL under Software, but it will be better informed, in any case, by considering your questions.


A long post here in response, cathyf, which if not read by others has some chance of your agreement, and more probably, your correction, rebuttal or refutation. I respond to your post here and, also to your post regarding a long slow decay of FR. I hope you sift through and consider this tome and can find something useful. Your critical response would illuminate and help my work here in Main and Software forums. (I blame you for provoking this, my longest droning on and my mulling over your interesting, well presented and illuminating dissection of your case.) 

In those two forums I am getting your and others' help--here's one long over due "Thank you" to you, cathyf--for help in my study of Resmed FL and its parallels to Tidal and Minute Volume reductions. Lately computer related trouble, which partly stems from this file-size-swamped keyboarder, has delayed further posting. (An aside here: I daily read most all about UARS and FL here. In doing so, it's puzzled me to see skepticism about FL, period, and about its connection to Tidal Volume drops. In keeping with your question, I ask you "What am I missing", besides brevity?)    

First, about your FR decays and FR resets. 

I believe those are largely (half? all?) explained by device, algorithm, and your bodily resolution settings or limits. And your and my posts re the 3 graphs cited tie-in with that, though not mentioned explicitly there as here. 

My thought about your FR decay with persistent FL levels is that continuous slight meanderings of FL--generic, ragged tip IFL actually, regardless of flags and scoring--up or down 1 or 2 ticks together with a definitional setting of a minimum FL duration prevent a breakout from the continuous FL pattern so it remains continuous. The other factor (other half?) explaining the FR decay, I further conjecture,  is your lifelong-trained physical/breathing constitution which powers through, maintains SpO2 levels, works hard and smoothly to ventilate down to your fatigue's or SpO2's critical point when your system alarm rings and causes a FR and FL reset. It's a two way "gradualism" by the device and your system, barring some other medical explanation. (It would not surprise me if you found other health issues of yours stem in part from lifelong SDB, but I digress again.) 

I think Geer1 makes a similar overall point to mine just above, but about overall limitations of device design.

Second, about your 3 graphs late in this thread.

I don't think there is anything mysterious about different FL flagging among the graphs. Resmed flags FL on the basis of a fuzzy logic sorting out of a combination of four factors with their pressure adjusting impacts on machine responsiveness (given a sound mask seal): instantaneous flow change and its variance from an abutting time window's average level, change in respiratory rate, an empirically established table of wave-shape adjustment factors and, probably, some kind of cardiogenic pattern detection and handling. IMO and in summary, the main trigger for flow limitation flagging is the machine/algorithm sensing of a local, here and now, change of flow rate.

I think a large part of our being puzzled by perceived Resmed FL flag inconsistencies arise from a patented and satisficing/balancing interplay between specific wave shape factors and flow-change caused reductions of TV and MV. 

The following--"nebulous as cloud and tea leaf readings"--little things come to mind regarding differences in character among your 3 graphs and their inspiratory flow limitation (IFL), cardiogenic effects, and FL:

I agree with Geer1 that our machines often seem unable to discriminate finely enough to infallibly call strikes (FL) vs. balls (non-FL). Part of that--my guess, correct or incorrect--is that the Resmeds react fast--sometimes prematurely--for pressure control and therapy purposes. They react to the incoming wave's flow drop or its offending wave shape signature as those most often portend increasing flow limitation or emerging apnea. The forming/just-formed wave volume (or its time at FR = volume or a surrogate) is compared to a preceding average from a number of waves. I think shape signature may have first priority to invoke a call for more pressure.

About your three graphs:

It looks like, and this may only be visual confirmation bias, that in the 1st and 2nd graphs there is a fairly "regular irregularity"--no large changes, just repetition of like irregularities-- in the two sets of inspiration wave shapes. But in the 3rd graph there are outliers amid the repetitive, accordingly there are FL. All three graphs show continuous inspiratory flow limitations (yes, it's "mmmm...") but only the bottom-graphed breathing shows marked changes among similar I-waves. As above, my sense of the FL flag is that it indicates when the machine/algorithm senses either a new, troublesome inspiratory wave deformity signature or an extraneous/outlier flow rate loss.


In the 1st graph I-wave at 23:19:55, that differing deformation is a more complex M-tip variant: the I-wave shape and area strike me as showing that the air volumes of its near predecessors was maintained sufficiently. Accordingly volume maintenance prevailed over the suddenly different shape that might otherwise have thrown a FL flag due to a drop in TV. On top of these considerations, confounding our interpretation, are the significant cardiogenic oscillations which may well be factored into the algorithm effecting a non-FL outcome in #1. (In the general case, respiratory rate variations--not evident nor considered here--are considered by the machine too.) It is similar to that more peer-deviant I-wave at  6:23:05 of #3. There was in its exhalation period no additional FL step up to be stacked on top of the persisting earlier FL level from I-wave at 6:23:00. Volume maintenance again overcame shape considerations--so no FL flag?

In the more irregular 3rd graph, the timing of the leading end of each FL appearance comes close to what I usually see, in falling somewhere near the mid-time of the exhalation after a particular deformed wave shape (lonely M-tip?)or sudden reduction in inspiratory wave area (TV). My frequently isolated M-tips trigger a flag more often than not and I think that is driven by the M shape, though its I-wave volume (TV) drops, maybe 11%, according to an analysis of one.

I tested my visual sense that not only was RR rate faster in #3, 10 breaths, vs the other graphs' 9 breaths, but I/E ratio and duty cycle were higher, but not as much as I expected. For sets of 5 sequential breaths, randomly chosen, these were averages of I/E, and of I-time/Total-time

I/E: ____________0.548, 0.636, 0.653 (2 high?, characteristic of IFL?)
I-time/Total-time:  0.262, 0.280, 0.283 ("duty cycle" fractions from a crude screen measurement device) 

It's amazing to me that, as I recall, you not only have low AHI, except for positional OA clusters (without your C-collar), but you have exceptionally long, flagged runs of unbroken and mostly high FL with little if any SpO2 drop. You somehow can power through and maintain mostly normal O2 level. Your FL presentations and flagging absences, as I see them, suggest that your negative esophageal pressures are--for you normal--exceptionally steady and highly negative during sleep. Moreover, you can generate the flow limitations at will if awake.

A further speculation: it seems that your continuous high FL suggests that your airflow constricting tissue, whether soft or more rigid, is and tends to stay more firm than is usual in sleep, not becoming slack enough to freqently spike your FL higher with near closures. 

Your case--unusal and instructive in my view--has intrigued me from the start, having once had untreated IFL, though having little now with my VAuto and C-Collar treatment. Along with our real SDB gurus here, I encourage you to get and try a VAuto out of pocket as I too found necessary for me to have one. Many find in it, as I did, their therapy answer.

With only a PS of 4 for my present used VAuto, titrated here by SleepRider, FL was and is cut down dramatically from my well titrated Autoset's much lower efficacy level using its max EPR 3. Only weeks ago, having found the CCWers' Belly Band hoster (to mount my accelerometer at Lumbar 3), I tried omitting the chock I had been wearing to prevent my preferred supine sleep. It took the accelerometer, registering sleep position and movement,to convince me I must prevent supinity. But my newer, higher C-collar (4 inch vs earlier 3 inch) prevents chin tucking that had to be the problem all along. Neither OA nor FL changed much, AHI is usually 0.0 - 0.5 now, as before, and FL are mostly small and scattered as I sleep on my back about half the night now.

I write all this and elsehere with much improved and more restful sleep. But it's been too many years since my best health and sleep to know if I ever did better, so I keep plugging away in that upward direction. Meanwhile, I hope to contribute helpfully while seeing so many deal with what once troubled me. Just gotta try to help and am grateful for AB, its experts, OSCAR and team, Mark Watkins and  all they have provided and taught me and others.


The image above, the link and the image below are edit additions inserted shortly after finding that the referenced image, as immediately above, had been removed and was not available in cathyf's post.

Also relevant here is matter on the breathing Duty Cycle. Here is a link to research about it as an indicator of hypoventilation, as well as a graphic from that paper.

Inspiratory duty cycle responses to flow limitation predict nocturnal hypoventilation | European Respiratory Society (ersjournals.com)


RE: Why don't these flow rate curves count as flow-limited? - cathyf - 06-30-2021

(06-28-2021, 11:36 PM)2SleepBetta Wrote: Also relevant here is matter on the breathing Duty Cycle. Here is a link to research about it as an indicator of hypoventilation, as well as a graphic from that paper.

Inspiratory duty cycle responses to flow limitation predict nocturnal hypoventilation | European Respiratory Society (ersjournals.com)

Wow-- I need to track down definitions of some of these terms (I don't know what "duty cycle" means but I think I'm following the gist of it). But if I'm following what that ERS article is saying, I think that they are talking about people like me!

Here's a picture of their graph of different FR curves on top (figure 2 from the paper) and below it I put in a very-typical-for-me 3 minutes of me in large extended flow limitation followed by an arousal. (I had no leak during that period). My machine has pressure pegged to 15/12 (I have EPR=3) and it edges down a bit during the arousal.


I think I even have a (half-baked) theory of how the EPR works for me. It allows me to have fast forceful expirations so that then the inspiratory phase can be longer and I can move more air over the extended period.

(If I'm right about how EPR works, it also really points up the utter cluelessness of sleep medicine in that EPR is on the patient menu not the clinician menu. It is way more powerful than setting the clock, and yet they treat it as a trifling trivial "comfort feature" that they ignore.)

(Also brings up another idea kicking around my brain for the last 7 years -- I'm a singer, and as anyone who has every sung in a group knows, 75% of choral direction is the director yelling "NO BREATH" in all those places you want to breathe, and "BREATH" in those few places the director wants to let you breathe. Maybe my breathing muscles are used to the concept of getting every molecule of air that I can when I can?)

RE: Why don't these flow rate curves count as flow-limited? - Geer1 - 06-30-2021

(06-30-2021, 09:37 AM)cathyf Wrote: (If I'm right about how EPR works, it also really points up the utter cluelessness of sleep medicine in that EPR is on the patient menu not the clinician menu. It is way more powerful than setting the clock, and yet they treat it as a trifling trivial "comfort feature" that they ignore.)

Kind of, you are half right.

EPR is limited to 3 which is what makes it a comfort setting instead of a therapeutic setting like PS on a bilevel (although 3 cm EPR can make a therapeutic difference in some people).The therapeutic nature of a bilevel is the reason that most doctors don't prescribe them readily and that you have to jump through hoops to get one through the medical/insurance system. You get a bilevel and higher PS levels once you have proven that a basic CPAP/APAP is not capable of treating your issues, sometimes getting the proof and/or convincing doctor that it is proof (especially in flow limitation cases) is the hardest part though. Often the only way to prove that APAP isn't enough is to get a titration study done that shows higher pressure support is required to overcome hypopneas and RERAs, unfortunately most people that end up getting a titration study end up with it being a waste of time because the tech spends majority of the time trying to first fix the problem by adjusting pressures rather than pressure support.

EPR/PS are extremely powerful for treating restrictive breathing. It not only makes exhalation easier by dropping pressure during exhale but it also makes inhalation easier by increasing pressure during inhalation. Pressure differential is the driving force behind flow, the higher the differential pressure the higher the flow. Higher flow overcomes flow restriction.

In your variety of threads many of us have told you that your flow limitations are real and that you should try a vauto with higher PS to treat them. You can try to go through the medical system and get a titration study etc that convinces doctor to prescribe bilevel but that will take time, effort and potentially money (if you have to pay some or all of titration study) and you might be better buying used.

RE: Why don't these flow rate curves count as flow-limited? - cathyf - 07-01-2021

I know I sound like Tevya in Fiddler On the Roof ("On the other hand...")

There is flat out no way I'm getting a bipap prescribed. I have severe positional apnea and can basically create a ridiculous AHI at will, so I have this diagnosis:

Quote:Impression: (The study was scored using AASM Hypopnea rule IA (3% oxygen desaturation/EEG Arousal)): ICD10 Diagnosis Code(s): OSA
1. Moderate obstructive sleep apnea, exacerbated to severe degree in REM sleep. Respiratory events were associated with oxygen desaturations down to a nadir of 86%. REM and supine sleep were captured during the study.
2. Abnormal sleep architecture likely due to respiratory events and first night effect

Recommendations and Plan:
1. Auto titrating CPAP 5- 15 cm of H2O with heated humidification and PAP download follow up is recommended.
2. Untreated OSA can cause substantial daytime sleepiness that may impair the ability to perform daily activities including operating heavy machinery or driving. The patient should be counseled regarding the risks of daytime sleepiness.

But as we know, "Auto titrating CPAP 5- 15 cm of H2O with heated humidification" is laughably ineffective against chin tucking. 

Without chin tucking I have virtually no events, and events is all these clowns care about. And even worse -- they only care about 30-day averages of events! And an hour or two of chin-tucking hot mess averaged together with a couple hundred hours of more or less zero is more or less zero.

Example: I had a night where I fell asleep reading without my collar on and had 47 minutes of positional apnea.
  • I spent 6:53 between first power-on and last power off. About an hour at the beginning was awake reading.
  • During the 47 minutes chin-tucking-palooza --
    -- I spent 20 minutes total time in apnea. -- I had a local time-period AHI of 65.
  • For the 6:53 "whole night" I had an AHI of 7.99.
  • MyAir took away one point for missing my 7 hour "sleep goal" by 7 minutes.
  • MyAir took away one point for the 7.99 AHI. One frigging point! for an AHI of 7.99!
(If I had spent another 10 minutes awake reading at the front end of the night my AHI would have been 7.80 and my MyAir score would have been 99!)

At the big-city specialty "sleep center" they "looked at my data" (i.e. they looked at Resmed's averaged-out lies about my data, not at my data itself) and said I was doing great! Oh, and they asked -- very politely -- was I going to keep changing my pressures or did I want them to take over that pesky little task? And here's the receptionist make an appointment to come back in one year...

But enough about those useless people.

Me? I run hot and cold trying to figure out if I have sleep-disordered breathing at all. Yes, I absolutely have really impressive positional apnea, but I can control that completely with a $12 cervical collar and don't have to use any machines for that. (And as we know, a CPAP is useless against positional apnea anyway.) Sure, I'm not normal with the flow limits that I produce, but one very real possibility is that I have atypical airway anatomy that I have adapted to well enough. (That's what the article 2SleepBetta linked to seems to be getting at. Normal people adapt their breathing to flow limits so that they get enough air.) Atypical rather than disordered. So I should put away the cpap and wear my $12 cervical collar to sleep in. Using the machine sets off a whole cascade of problems to solve. I can't seem to get the leaks under control with the nasal masks, and the cervical collar pops the magnets off the full-face mask, so I can't get leak-free with that, either. When I set the machine at 7-15, EPR3, I really don't have that much in the way of flow limitations during periods of no leak, or at least not much for me. But I can't figure out whether the flow limitations that I have during leaks are some artifact of the leak (the machine isn't properly accounting for the air leaving the system) and as such become no problem without a mask to leak, or whether the flow limitations come first and they cause the leaks, and without the mask I've got untreated flow limitations. My Fitbit detects some of the arousals, but it only has a resolution of 30 seconds, so an arousal under 15 seconds gets ignored. I KNOW that the arousals kill me, and the mask leaks and the raw bleeding sores I've got going under my nose aren't helping me stay asleep...

And the nearest vauto I can find on craigslist is an 8 hour drive away. Taking EPR from 0 to 3 has made a huge difference in my detected flow limits, but I have been getting a small number of centrals popping up. I'm going to be p**sed at life even more if I drive 16 hours round-trip to pay for a vauto out of pocket and end up finding out that pressure support of 3 is my upper limit and above that I get too many centrals! (Everything else in my life the last year has gone that way; why should this be any different? Rolleyes ) Add to that the fact that  now there are how many millions of Phillips patients who really need that vauto more than me.

The professionals told me last November that I don't have apnea and I should go away and leave them alone. Then after I did a chin-tucking hot mess of a second sleep study they gave me a basically useless plan and an argument that I'm too stupid to manage my own settings. All while everything that they tell me about my sleep compared to the data that I can see for myself makes it  abundantly clear that I may be an idiot but they are way stupider than me!

RE: Why don't these flow rate curves count as flow-limited? - 2SleepBetta - 07-01-2021

Don't remember it accurately but laughed long at Tevya's looking upward to God and asking if it really would mess up His plans if his life went a little easier for him on earth. Great show.

I understand the frustration and am likewise amazed at poor Sleep MD treatment you and I got and we see reported here regularly. I hit 5 years with my Autoset last September, DME started pushing to send me another. I've held them off wanting to go for a new VAuto. But my alleged best teaching pulmonologist (new to me then, soon advised against getting a Bi-level in 2018 or '19, probably because he assumed I was too stupid and would set pressure high and get CA's, trying to cut FL). He came around later after seeing my used VAuto's SD card data. Bigger point to make, is we soon worked well together, little we do, but he tells me it is unlikely I can get a VAuto with, effectively fullly treated apnea, and only a low level of residual FL (the VAuto achieves). Of course, I could generate 30 days of dirty FL with my old Autoset. Further, to try for VAuto would take another sleep study, he is sure, which I note, of course, would put overall update cost well above insurance just paying more for VAuto than another Autoset. At this stage, undecided, I'd like to know if there are any significant changes/improvements in the 11 series. Is it just a new touch screen and cosmetic differences, no functional differences, and wlthout any reduction in my data access nor freedom to do all settings? No matter what, I'd have to make it a used machine by using it to "comply" for at least 70% .....etc. So would not have a truly unused machine to sell to maximally offset cost of a new VAuto after the time I came into ownership of a new 11 series. I have Autoset backups.

I've had good luck with OfferUp for things in low $100's. You get two days to inspect your goods and, OU promises you can get full refund if items are not as offered (I say as specified). For example, my VAuto, $300 a couple years ago: Just before offering an amount I used the Ask Question feature and stated purchase conditions to the offeror: run hours limit, non-smoking environment history, time limit of the offer I promised would follow immediately, other device conditions of soundness, cosmetics, contamination, smell and such--don't remember which or all now. BUT, I copied every exchange with seller keeping a full screen shot. Anyway, I now see fewer VAutos, but there are ones in CO, CA and elsewhere at Offerup, $400-800. You have to search all permutations of "CPAP, BiPap, Resmed, VAuto" and combos of them. Just a suggestion that you consider OfferUp (where I probably could get burned somehow next time for the first time). The site lays out how to get a refund (a bit involved).

Obviously I'm not a doctor, and have only guessed at what assessments I've made of what you present. In the image below  I did the best I could with your small (lesss than photo sharp) image, rough tools and shaky hands to show you your Duty Cycle (= Time to inhale/(Time to inhale + Time to exhale) for one breath and its I/E equivalent look.  DC, fairly new to me (except for old stick arc welders) is often used and is stated as a level, range or condition benchmark. In the ERS paper they show, for the small study, pre-dosing "normals" had DC, men and women, of 0.40, as I recall. But when they were dosed with air deprivation the severe FL level was 0.51. In your graphic, the lowest DC was 0.50 and there was only one that low  in the lead up to the RERA. The  high DC was  0.76 among many 0.70 or larger, as marked with green dots. The I/E ranged 1.0 to 3.2. Numbers by wave were measured time units. Cardiogenic effect detracts from accuracy, so I tried to guesstimate a compromise where they were worst so as to measure real parts of an I-wave.

The lower Red dots were marked first by only visual inspection--missed a couple of the 0.70's. Then "A" and "B" were inserted trying to show B identifying smaller TV waves than A, the only "squashing" effect visible in the 3-minute view. I later decided to try to show the DC and I/E times.

The upper red dots showed visually obvious I/E's >1.0. The main feature is the FL squashing of amplitude (~TV), for one experimental individual with his baseline and deprivation doses.

The tables and figures in the ERS piece are worth reading. I had to jot down what the many subscripted elements  were to think straight about them though.

As I mentioned, you may-be/seem somewhat inured to, and be able to continue to ventilate after lifelong flow limitations, but I believe it is not without high insidious cost to your health and sleep as my severe OSA was to me, though I was  not obviously affected with tiredness. We should rest in sleep not work harder and harder to breathe or to start breathing again after stopping.

One  more thing about my (1) long time focus on wiggles in peaks of I-waves vs. flow limitations and (2) about working on finding a way  to use FR data file info to calculate and score the near-actual missing part of breaths lost to flow limitations. Selfishly, its been worthwhile (educational and interesting) for those efforts, to follow your presentations--your condition, your comments and your documentation.