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

[2SleepBetta]

@sheepless:

The OSCAR kind of programming is a separate world I know little to nothing about. 

To do what I suggested is childishly simple with Excel--once one has the PLD file converted to txt file(s?), thanks to the EDF browser either Crimson Nape or pholynyk suggested. The desired total would appear if a simple SUM function were entered in one cell to sum the FL column, one column of about 10 in the PLD file. To the extent sleep data for a night, hence my "?" mark, are in a number of partial data files, the sums from each would need to be totaled. 

Aside from FL distribution info, sums will have neither more nor less meaning than the individual FL flags have.


The very simple EDF conversion step will generate a number of files, from which only the PLD data file(s?) will be needed, the rest can be deleted if not needed for other work.

[cathyf]

I'm still stuck on the point that my machine clearly misses a lot of flow limits, and so any grand totalling is going to be heavily skewed by how well the machine happened to do at seeing what was there.

The main characteristic of all of us with high flow limits is the spiky appearance of the curve, where it's pretty rare to get enough non-zero data points in a row so that a whole-night view shows any hump rather than just vertical lines. Normally you have to zoom to see any daylight underneath that graph. I maintain that there are four completely different meanings to each recording of a zero flow limit. The first meaning of zero is that the breathing is not limited. The second possibility is that the machine was doing something else and didn't calculate a value at all. The third meaning is that the breathing is so wildly jagged that no calculation is possible. And the dumbest one is that the machine records zero for FLs during OAs -- isn't a complete obstruction the very essence of a total limit? That last one you could tweak pretty simply, in that you can define an OA as a flow limit of 1.0, and add that into your index after totalling up the reported FL values.

The fact that the median FL is usually 0.00 or maybe 0.01 even when the graph is densely colored with tall spikes tells you that you've got a huge amount of the night where the machine isn't paying attention to FLs at all. You've got to account for that somehow

[2SleepBetta]

@cathyf, your comments in black, mine in blue:

Notes:

1.  My Autoset, manufactured in 2015, could differ a bit or a lot from yours on FL flagging, because of our YOM and run hours differences.
2.  I'll be referring to my graphs of your 6.5 hours of 6/19/21 sleep data in my nearest graphic above.
3.  Three things to keep in mind about the red and rust TVd colors  and the yellow and rust colors in the graphic: 
     (a) All solid red and rust color other than for inverted ceiling grass taller than 0.10 should not be there, because your ruler-eyeballed average TV is 0.4 L approximately (0.4 L) should have been the graphic's illustrative base line--my bad--not the 0.5 L baseline which causes "TVd = 0.5L - TV "to be overstated 0.10 L; 
     (b) All our TV varies, accordingly TV-relative TVd varies; and 
     (c) Apparent differences and equality of y-dimensions (for red, yellow or rust traces) is a scaling accident, having no meaning beyond the visible fact that both the red or rust TVd and the yellow or rust FL values rise or fall concurrently but not equally.
4. Machines perform a bit differently at times--arising from built in differences in tolerances, wear, heat, atmospheric pressure, humidity, line voltage, and in the good old USA of days gone by, the day of the week it was built, etc.


I'm still stuck on the point that my machine clearly misses a lot of flow limits, and so any grand totalling is going to be heavily skewed by how well the machine happened to do at seeing what was there.

In my case and history that would not detract from beneficial use of the grand total as crude summary of what I see distributed through sleep. Further, and not disputing you, I don't see that in your first 6.5 hr June 19, 2021 graphic as above. Not once a 0.10 L-wide red ribbon is visualized as snipped off your red or rust TVd along the 0-axis and all remaining red grass TVd is visualized as reattached to the axis. There will be little disagreement on whether there are FL and whether it is increasing or decreasing flow limitation in either the RM or TVd sense. A different night with different cathyf airways, etc., then different FR, FL and initial conditions for the algorithm and detectors to sort out that night. As we know, certain deformed wave shapes among "normals" may be flagged one session/time-segment, not the next, contiguous/continuous/"bridging" FLG one night will be separated the next night. Other FR, pressure or FL context enters in in some mysterious way as the devices set about their main job adjusting pressures, yielding FL flags as a machine control byproduct.

Would it be better for you if there were no FLG? As I recall, you or others have suggested review of the FR curve shows sufficient info obviating much else.  

The main characteristic of all of us with high flow limits is the spiky appearance of the curve, where it's pretty rare to get enough non-zero data points in a row so that a whole-night view shows any hump rather than just vertical lines. Normally you have to zoom to see any daylight underneath that graph. I maintain that there are four completely different meanings to each recording of a zero flow limit. The first meaning of zero is that the breathing is not limited. The second possibility is that the machine was doing something else 
and didn't calculate a value at all. The third meaning is that the breathing is so wildly jagged that no calculation is possible. And the dumbest one is that the machine records zero for FLs during OAs -- isn't a complete obstruction the very essence of a total limit? That last one you could tweak pretty simply, in that you can define an OA as a flow limit of 1.0, and add that into your index after totalling up the reported FL values.

True, no TVd, no FLG for that, FLG=0. But no FLG also means that in present look back context there is insufficient basis for FL flagging of a wave shape, a wave flattening nor a duty cycle change--any single one of which, or combination of them, would "merit" a FLG in a different context. Failure to change pressure appropriately for wildly jagged breathing is a weakness for APAP. It would be desirable, I think, to have the device switch to CPAP, for a time, once its hunting up and down could be detected after a few exhalations. Re 1.0 for FL at apnea, fine, but redundant and not hard (for me) to accept one way or the other.

The fact that the median FL is usually 0.00 or maybe 0.01 even when the graph is densely colored with tall spikes tells you that you've got a huge amount of the night where the machine isn't paying attention to FLs at all. You've got to account for that somehow

Memory may fail here, but I believe AB techs have acknowledged that one or both the 0.00 and median FL summaries are wrong or deficient and to be reviewed and corrected at some point.

[cathyf]

Memory may fail here, but I believe AB techs have acknowledged that one or both the 0.00 and median FL summaries are wrong or deficient and to be reviewed and corrected at some point.

A quick aside here -- I think that the issue is that the column is labeled "MAX" but it's really the 99th percentile. I think that the 99th percentile is the number you really want here, so we should change the label not the calculation. On the other end, I think we should almost always be using the 1st percentile value rather than the Min for most everything that we calculate in that Statistics table -- it filters out a lot of garbage that way.

[Crimson Nape]

cathyf - If you are referring to OSCAR, then in the new release, version 1.3.0, the Max label has been replaced with 99.5%.  If you are not referring to OSCAR, please ignore this post!  

[cathyf]

To be hopelessly pedantic, it probably should be "95th" and "99th" or "99.5th" rather than using a "%" character. There obviously isn't enough room for even "95th%-tile" let alone "99.5th%-tile" so some compromise needs to happen (or really teeny-tiny type LOL)

But, anyway, it doesn't matter so much...

[2SleepBetta]

First and off this post's topic: (1) There is an error in the heading of the lower graph in the attachment to my post #39 above. That graphic is here attachment.php (1204×773) (apneaboard.com) . Where that lowest heading begins it should show "Ti/(Ti+Te)" or "Ti/Ttotal" instead of "Ti/Te". (2) In later work I saw that the red TVd (tidal volume drop) indicator bars should have been 0.10 L shorter.  That is because the sleeper's average tidal volume, as can be seen graphed, was very close to 0.4 L instead of the commonly cited 0.5 L average value. Accordingly the red band of TVd should be shrunk upward by 0.10 L as posted elsewhere. The corrected formula would be TVd = (0.40 L) minus (measured TV) not TVd = 0.50 L - TV .

Detail in my avatar shows the scheme: The area bounded by one inspiratory flow rate curve and the axis is the TV and its volume is represented by the height of the green vertical bar. The TVd from an (off graphic) flow limit is represented by the hanging red bar. The red and green bars usually add up to near 0.50 L as noted above.


Now to this post:

The attached OSCAR screen shot below shows the same graphic I repeatedly used in my "Call for Excel help" thread in the Software Forum. It is an attempt to show an approach anyone, particularly FL nerds like me, can use to assess our long duration and varying Resmed FL flags or our dense series of them. Only an OSCAR presentation, a ruler, a simple calculator and a desire to know are  needed.

As in the graphic linked above, we see that FL flags and tidal volume (TV) losses/drops (TVd) agree fairly well on the presence of flow limitations, but they often seem to widely disagree on the severity or seriousness of them. That is because a FL flag can indicate any one or more of  four different things, including TVd, but may not reflect any TVd at all in a FL flag.

Our AB experts have shown us that one of the best ways to assess seriousness or extent of flow limitation (whether it is flagged or not), is to look at whether inspiration time is increased. As they know, better but slower measurements are the I/E and duty cycle (Ti/Ttotal ) ratios.

One way to spot high ratio periods is to put the Inspiratory time curve above the Expiration time curve and look at flow limited areas where the two curves diverge widely. It's not fool proof because the I-times can be shorter along with shorter E-times and have a high I/E or duty cycle ratio when the two curves are not widely divergent.

On the graphic I note that an increased duty cycle is preferable to an increased respiratory rate when dealing with flow limitation. That's as if we have a choice of which to use in sleep. Not so unless there is some kind of breathing training  (Buteyko's methods?) that we could do to shift away from RR toward a duty cycle response.

Additions and corrections are welcome, as always. I zig zag trying to analyze, learn and present these things as helps for me and discussion. 


   

[quiescence at last]

So, I am trying to understand what caused the immediate and complete shift from short insp, long exp to even insp and exp. I see things like that happen in smaller periods in my charts, but they shift in, then back out a few minutes later.

QAL

[2SleepBetta]

@QAL:



Your question prompted me to revisit that OSCAR profile and sleep session and to look at what I had written about those changes in another thread at post no. 23, here Call for Excel VBA help: to support effort to clarify/understand FL | Apnea Board . That account is about the best I can offer. 

What I can add or emphasize is that this session was in my 10th week of PAP when therapy was already much improved-- thanks to pressure changes I made after lurking here (after my rough start 9/11/2015 as shown in part here attachment.php (1294×593) (apneaboard.com) . 

-I would not have been wearing a C-collar nor using my supine-block, both of which I greatly benefitted from within about 1 to 2 years later.

--FL sleep before the OA cluster had rounded inspiratory wave tips and, I'm quite sure, had to be on my left side.

--OA had to be when supine ('never have any of those anymore, 2 of them >50 seconds). 

-HIgh numbers of "M" inspiratory tips are clearest indicators I was on my right side and experiencing far more flow limitation soon after the OA. Guessing, I must have tucked my chin a lot as soon as I turned to my right side. On my right side cardio ballistic waves are worst and that likely accounts in part  for the more erratic I and E time curves.

--Main point : The increased flow limitations after the OA caused need for much greater inspiratory efforts--higher I/E and  duty cycle ratios to maintain TV. Accordingly, markedly longer inspiration times and such higher ratios are good indicators there are flow limitations, flagged or not. Without extra inspiratory effort tidal volume would drop.


Nearly the bottom half of this linked post is my account of my related recent surprise. It underscored need for high enough C-collars--how those can do wonders for the OA troubled: Can somebody talk me into a nasal mask? | Apnea Board .

[quiescence at last]

Thanks. I will be rereading some of those posts, and keeping better track of periods I am on my right side. That is pretty rare, as it faces toward another sleeper, blowing air toward their face. I would be more likely to tuck my chin in that case, to redirect air.

QAL