Hello Guest, Welcome to Apnea Board !
As a guest, you are limited to certain areas of the board and there are some features you can't use.
To post a message, you must create a free account using a valid email address.

or Create an Account


New Posts   Today's Posts

Call for Excel VBA help: to support effort to clarify/understand FL
#41
RE: Call for Excel VBA help: to support effort to clarify/understand FL
Note: The following matter, focused on Resmed FL flags as this entire thread is, does not address any UARS flow restrictions which may only manifest/present in labored breathing, abnormally high I/E ratios and/or inspiratory wave tip deformations that are not flagged with FL markers.  The lattter two can be seen in OSCAR curves. But the markers of the more subtle sleep impairing UARS must be discerned from other and additional signs, as is most accurately done by a sleep lab employing a Pes measurement device or chest-effort  strap and the data those provide. That is not to say that what is discussed below is irrelevant to a UARS sufferer, because I don't believe there is a bright line between UARS and other flow limited breathing  as flagged by Resmed devices. "It" is a gradient thing.

@sheepless

    1. Re the 4-second steps:  

In my 6 years of Resmed-PAP-reading experience, I don't recall seeing any FL of any severity having a duration less than 4 seconds. I believe the sampling rate is 0.5 Hz, every 2 seconds, one factor involved here.

The algorithm apparently is set to a 4-second minimum display, not coincidently it seems, because at a RR of 15 each ventilation cycle is 4 seconds. There is no other connection to the variably spaced TV bars and their coinciding TVd bars (if any), which are computed and displayed at the precise end of an inspiration having variable duration.

    2. Re how duration is included in the OSCAR percentile summary:

I'm on shaky ground here, never having gotten it crystal clear and locked in memory just what has been said by AB tech people about the OSCAR FL-percentile summary. That said, my intuition is based on the fact that the PLD file column with FL severity values presents strings of numbers  (2 seconds apart as I recall) as follow:  ........0.0, 0.1, 0.1, 0.2, 0.2, 0.2, 0.5, 0.8, 0.6, 0.3, 0.1, 0.0, 0.0, 0.0, 0.2, 0.2, 0.2, 0.0, 0.0, 0.0, 0.0, 0.0 ....

Both of the first two non-zero values will probably show as 0.1 severity. The remainder of that first FL string will look like progressive stacking and unstacking in the graph. The series of "2s" will show as one FL with severity 0.2 which  will persist for up to 12 (= 3 x 4 ?) seconds and, in any case I believe, have a duration that is divisible by 2. How those strings add up and end is unclear. For example, I just looked at a stacked FL of 0.13 max severity which level persisted for 6 seconds. All FL do start and end on a whole "second mark".

    3. Percentile scoring:

My intuition is that the percentile score is based on the percentage of seconds (or 2 or 4 second intervals)  in  the sleep session being divided into size-ordered (increasing or decreasing sequence) counts of individual severity levels. The divisor and the weighting of each single severity value must be for the same number of seconds.

In conclusion, if my intuition is  correct, then both duration and severity are represented, as above, in, say, the 95th percentile (or whatever OSCAR's accurate percentile number is in the Summary).

    4.  "Stacking":

It only looks like stacking, stairs or layering. The look comes from the severity numbers increasing or decreasing as indicated above.

     5. My sense of the Resmed FL domain in general (noting that the fuzzy logic Resmed declares in its patents blurs many "edges" of my outsider's attempt at definition):

With a FL detection and flag the device algorithm is signaling any non-apnea inspiratory irregularity--as such irregular data can be viewed in FR curves--from a flow limitation just at or above the flagging threshold (in a then-prevailing I-tip deformation "climate" it has sensed) and ranging up to full a scale actual airflow limitation of 1.0 or close to it. 
If FL=1.0, the TV would be zero or very nearly so and the breathing stoppage shorter than that for any form of scoreable apnea. 

The FL warns of breathing interference of some kind on the basis of its detection of one or more troublesome tabulated or sensed wave forms, a RR disturbance or a TV drop. 

     6. In summary: 

Again, my sense is that the sum of the flag areas (i.e. duration times time) and how the total varies sleep session to sleep session could be helpful. Helpful in the same sense bars showing the spectrum green through red are used to indicate/suggest overall risk, contingent danger, hazard, etc., when one or more trouble and/or beneficial factors are involved. It's "the view from 30,000 feet", all things considered. Nevertheless, using another meme (?), the devil's in the details, which should be reviewed in some detail when there are notable changes from what has been understood and seen as normal.
I have no particular qualifications or expertise with respect to the apnea/cpap/sleep related content of my posts beyond my own user experiences and what I've learned from others on this site. Each of us bears the burden of evaluating the validity and applicability of what we read here before acting on it.  (Disclaimer use permitted by sheepless)

 
Post Reply Post Reply






Donate to Apnea Board  
#42
RE: Call for Excel VBA help: to support effort to clarify/understand FL
Here's another long post as I  attempt to review in some detail how I believe work in this thread has added to understanding of my flow limitations above and below the Resmed (RM) lower threshold for flow limitation flagging (FL). For now I am uncertain how to proceed unless I or someone else simplifies drilling down into and using easier methods than Excel spreadsheets for exploitation of the many detailed and disjointed data files Resmed (RM) smart devices create during our sleep. 

An opening reminder:  All my analytical work here has been and is anecdotal. It is all from one single bimodal (low and then high FL) night of what I think was my most richly varied SDB night near the beginning of my therapy. But I do believe my zig zagging toward understanding the FL flag here has clarified the meanings of those FL to some extent. But now I've hit a brick wall of patented algorithm secrecy where only plausible conclusions can be offered from analysis of and thinking about RM data graphs. Of course, numerous refinements can be done--in addition to corrections and clarifications. (I'd like to see graphs of moving averages of Tidal Volume [TV] groups and their drops [TVd] and how those compare to TV we see in OSCAR. Likewise, it would scratch my curiosity itch to see breath by breath and moving averages of the quotients of Inspiratory and Expiratory time [I-time and E-time].) 

First, an open question comes to mind begging closer reading of the Resmed patents, word from experts at this site and/or revisiting charts that often show FL amid continuously adequate TV levels. From a therapeutic point of view, is it not more important that  SDB patients know when their  ventilation (their Tidal Volume [TV] half breaths) have fallen below their healthy levels appropriate to their  weight, height,  sex and need for restful sleep  (all but the last being ICU ventilator setting criteria) than it is to know their TV has fallen from some higher to  lower level but is, at its lowest, still fully adequate ventilation?

As posted earlier, I assumed Tidal Volume drops(TVds)  that do not reduce TV below acceptable fixed minimum levels are irrelevant, just as I assume TV greater than locally quiet sleep need is (except, that is, as may be needed to recover from a brief TV deficit--recovery breaths). Only TV deficiency is a concern. But can variable or highly variable TV above the level of necessary TV be entirely innocuous, innocuous for those who present only that variability as a single possible sign of disease? 

My post no. 10 above noted an anomaly I've come to realize showed only a simple failure to understand  how all metrics fit together in the seemingly valid approach I'd taken. (An attached graphic addresses this matter.) The seeming anomaly was during a period of high FL when my graphs of TV drops showed few and no drops, no TVd for short periods. That was simply  because (duh! now the light comes on) actual TV for a time exceeded my assumed TV need of 0.5 L from which I had deducted actual TV to determine TVd. The vanished TVd was the result of "extra" actual TV > 0.5 L "washing out" all TVd, as the graphs revealed. 

Professionally published normal breathing TV ranges from 6 ml/kg to 8ml/kg  or, very approximately,  about 500 ml is usually cited. My VAuto currently shows mine as 480, 500, or 520 ml night long. The full range of TV need details can be found here, as can be seen elsewhere, by clicking the "Predicted Body Weight Calculator" link at: http://www.ardsnet.org/tools.shtml . Using my VAuto's 500 ml shows my spot in the TV range cited is at 6.25-6.5 ml/kg. 

But trouble is,  normal TVs are not constant through the night. That presents a dilemma. RM addresses this, as I understand now, with a FL flag marking the TVd from a recent average TV level for a number of inspiratory waves (IW.) My simpler (but hard and tedious to apply) approach ducks RM's complex burden of determining the moving TV average in my use of an approximate constant value, TVs of  0.48, 0.50 or 0.52 L.                     

My point, which may well be wrong,  is that except in extreme pathological breathing cases adequate normal breathing will vary enough to cause some RM flow limitation flagging, "FL".  If we know our acceptable minimum  TV level we can check our TV and MV curves and our  Inspiration to Expiration time ratios for breathing adequacy and dismiss innocuous FL and TV drops. The question remains:  is significant  ventilation variability bad at any and all TV levels?  As I hope I recall  correctly, the Sao Paulo SDB study showed that about 30 percent of untroubled, normal-breather- sleepers have a significant level of flow limitations without experiencing or showing evidence of poor sleep.

Onward to details of what I think I know more accurately. These are about flow limitations flagged by RM devices as FL:

1. A significantly reduced airflow, a drop in FR, will be flagged. But a constriction of airflow that comes from constant constriction of the airway and is mostly continuous will not usually be flagged. The latter causes more negative esophageal pressures and increased (compensating) work in drawing breaths giving rise to higher Inspiration times. The latter  may be discerned from deformations of the IW tip (dIWT) and other metrics RM devices provide. Those are graphed in OSCAR as Inspiration time and Expiration time. Either an Increased I/E ratio or an increased duty cycle, I/(I+E) are metrics used to describe this in sleep literature . 

2. Respiratory rate is figured-in by the RM algorithm in its flagging. How? I still have no idea unless it is an element "buried" in the duty cycle or, similarly, in the I/E time ratio. Some persons experiencing airway constriction will compensate with longer breaths with longer I-times and shorter E-times. Others will present higher respiratory rates. Either response is to increase and maintain needed ventilation,  but the slower deeper breaths are more effective ventilation because TV loss, dead space ventilation loss, is lower at lower RR if sufferers can breath deeper and maintain a lower RR.

3. One or more significantly  flattened inspiration curves will almost always be flagged.

4.  Other  wave shape irregularities--including two particular ones among them, the "M" tip and the "Chair" tip: Even a single slightly M-tipped inspiration wave (M-IWT) often causes a low severity 4-second FL flag, similarly that is true for some singular sigh-wave tips. The latter are asymmetrical  M-IWTs  that resemble the frontal view of a rabbit's head with its two ears, of course, but one ear is lopped-down.

Here is a  very murky FL interpretation  zone: one involving those common lengthy series of  similarly deformed and flagged dIWT. But note that the same wave train on another night or period of the night may not be flagged.
 
A series of M-tipped, Chair-tipped or other regularly occurring and somewhat misshapen dIWTs will sometimes--apparently when their detectability hovers near sensitivity or resolution criteria/limits  of the Resmed algorithm--  result in a broken series of separate small FL. 

But at other times, those misshapen dIWTs will seem to begin laying down continual contiguous FL layers that looks like a stacking of them upward, this as an all-"alike"-dWIT series often lengthens for long periods (minutes?). The  reality is that the FL severity value number increments upward one notch (0.01  or more) with each successive dIWT until a peak is reached and a drop ensues at some airflow or pressure change--for whatever reason. 

The first dIWT of the series will seem to lay and continue (?)  maintaining the bottom course of FL, the next wave of the series will lay and continue (?) a second course, the next (the third wave) will lay and continue, etc., with FL "severity" (i.e., by virtue of stack height) seeming to grow larger, but the visually detectable dIWt shape and TV value seem invariant. 

FL Severity is represented by "stack height" value according to RM and will overstated for the individual dIWTs in a lengthy series, but on the other hand, if a long duration low severity series of  irregularities (a long series of dIWT) has been flagged, it might (by MDs, that is) be deemed more serious and notable than scattered minimal FL. That latter evidence based judgment, if extant and plausible, would add a severity dimension for duration itself. Accordingly, all FL could/would have two dimensional severity along with a distinct duration dimension. 

See attachment.php (1497×920) (apneaboard.com)

At some point a change in dIWT forms or another FL criterion will stop the algorithm's stacking or stair climbing and , most usually, a "stairway" of FL steps downward will be taken--faster and steeper downward than they went upward. As always, a significant flow rate change triggers a RM device response which we will see in one or more OSCAR curves.

The beginning of a stair laying or stacking process is easy to see and understand, but long runs of stair stepping or platform maintenance, it seems, can and  do absorb and incorporate new  series of FL signals from later dIWT which, when those series of moments arise, will  variably maintain the pre-existing severity platform and may build onto it until the elevated severity platform level will drop because of changes of FL conditions.  Sometimes it seems the FL platform has a tendency to seem "sticky" upward. It may be the algorithm maintains a longer term look-back register that resets the algorithm perspective and perceptions--part of the "fuzzy logic" RM's use.

Note: Neither RM nor its patents directly reveal all of (any of?) the significant control values that are  embedded in and used by RM device algorithms. But our analysis of RM curve data yields some useful inferences about significant values. For that it is back to "reading entrails", zig zagging and bumping into guard rails--hopefully not busting through them as I've droned on and in this thread.

Any readers' corrections, constructive criticisms and questions are welcome as always. 

A separate thread of mine will link back to his thread just as I link to it below now. Both it and this thread are focused on flow limitations and have some worthwhile links and graphics for those troubled by flow limitations to consider. Further, as I believe I have mentioned elsewhere, there is the difficult AB case of flow limitations presented by member cathyf (in her earlier FL-focused posts particularly) and the question about summarization of nightly FL raised by member sheepless:  both probe deeply into the problem of flow limitations and include many posts by Apnea Board experts and astute members. Those two recent discussion threads come to mind, but using Google to search the AB site I am sure much else of value can easily be found posted at AB as well as a wealth of key information in the Apnea Board Wiki.


Upper airway resistance syndrome (UARS) - Apnea Board Wiki


AHI < 1.0 now, but still tired? Is it UARS, RERA, arousals or FL in your FR peaks? | Apnea Board


flow limitations expressed as an index? | Apnea Board


Call for Excel VBA help: to support effort to clarify/understand FL | Apnea Board

attachment.php (2550×1650) (apneaboard.com)


attachment.php (1650×1275) (apneaboard.com)

   
I have no particular qualifications or expertise with respect to the apnea/cpap/sleep related content of my posts beyond my own user experiences and what I've learned from others on this site. Each of us bears the burden of evaluating the validity and applicability of what we read here before acting on it.  (Disclaimer use permitted by sheepless)

 
Post Reply Post Reply


Possibly Related Threads...
Thread Author Replies Views Last Post
  OSCAR Support for DSPD (Time/Session Issue greenwsj 7 232 09-02-2021, 02:45 PM
Last Post: fakename1845
Smile Oscar support for ResMed AirSense 11 Autoset tmiller 4 486 08-23-2021, 05:08 AM
Last Post: tmiller
  RESSCAN - PURE VIRTUAL FUNCTION CALL [see post #2 for interim solution] srlevine1 4 175 07-29-2021, 09:59 AM
Last Post: srlevine1
  ResMed iVAPS support? alshayed 4 142 07-12-2021, 04:46 PM
Last Post: alshayed
  Support for Fisher & Paykel ICON+ Auto on SleepyHead or Oscar JaX F 2 269 04-05-2021, 12:17 PM
Last Post: A KLERK
  Help with Excel Correlation please SevereApnea 5 298 03-29-2021, 09:26 AM
Last Post: Crimson Nape
  Adding Daylight Savings support to OSCAR greyham 16 751 03-25-2021, 05:23 PM
Last Post: kappa


New Posts   Today's Posts






About Apnea Board

Apnea Board is an educational web site designed to empower Sleep Apnea patients.