False Flag Events

I've been using a Devilbiss/Intellipap APAP machine for ten weeks now and have found this forum very helpful, not least because of the many links here to information sources on how to interpret sleep data from Oscar. I'm fairly certain that about 50% of the flagged OA/H events on my charts are false. I get a few centrals also (maybe 3-4) and these all appear to be real. As a result, my AHI (averaging around 2.5) is probably overstated by about 80%, i.e. it's probably less than 1.5. Just wondering if this is unique to my machine or if it's common to all machines. Here are two examples from last night. The first one is a flagged OA when I removed the mask briefly to have a sip of water. And the second was a flagged 20 second hypopnea, which, at most, was some short duration flow limitation as I see it. There were a lot of other "dodgy" hypopneas too, many of them because they were obviously less than ten seconds.

I suspect the pins don't drop exactly in the place where you might expect to see them.

Re the OA: it looks like you held your breath just before you took off your mask. Was the breath-holding 10 seconds long? If so, that might be what triggered the flag.

Re the H: this looks like an hypopnea -- the period of reduced flow and then what might be a tiny arousal afterwards. Here's the current definition of hypopnea: ≥ 50% drop in flow associated with a ≥ 3% decrease in saturation or an arousal.


Of course, the machine has no way of determining what is happening with your O2 levels, nor does it score arousals, so in that sense it can't really tell you whether you've had hypopneas.

All CPAP machines have some degree of false flagging. The machine only knows about flow and pressure, but not why the flow is irregular. For example, if you hold your breath for 10 seconds (as you might do when turning over), the machine only knows you aren't breathing so might flag that as an apnea. In this case, the flow graph might show an arousal before the apnea rather than after.

OSCAR shows what the CPAP machine reports, so there should not be a discrepancy between OSCAR and the machine.

It is always possible that OSCAR is not displaying the flag in exactly the right spot in the flow graph. This can happen when the CPAP machine reports event times with a different resolution than it uses for the flow graph. I haven't looked at the DeVilbiss loader in some months so don't recall how closely the event flags displayed should match the flow graph.

Miss Emerita wrote: ↑

Fri Nov 26, 2021 1:05 pm

I suspect the pins don't drop exactly in the place where you might expect to see them.

Re the OA: it looks like you held your breath just before you took off your mask. Was the breath-holding 10 seconds long? If so, that might be what triggered the flag.

Re the H: this looks like an hypopnea -- the period of reduced flow and then what might be a tiny arousal afterwards. Here's the current definition of hypopnea: ≥ 50% drop in flow associated with a ≥ 3% decrease in saturation or an arousal.


Of course, the machine has no way of determining what is happening with your O2 levels, nor does it score arousals, so in that sense it can't really tell you whethetr you've had hypopneas.

Yes, I know, there's about a fifteen second delay.

With the fifteen second delay and because the flagged OA was 25 seconds long, it appears to have mistaken the time when I removed the mask and paused the machine as an OA.

The increased breathing afterwards indicates that my airway was somewhat blocked for a period beforehand (the hypopnea). But to me, looking at the area above the zero line, it's unlikely to be less than 50% of my normal breathing unless the period looked at is less than ten seconds. So I still reckon a very marginal call. Maybe it might be worth my while to invest in a pulse oximeter with blue tooth capability? Thanks for the feedback.

BlueDragon wrote: ↑

Fri Nov 26, 2021 1:19 pm

All CPAP machines have some degree of false flagging. The machine only knows about flow and pressure, but not why the flow is irregular. For example, if you hold your breath for 10 seconds (as you might do when turning over), the machine only knows you aren't breathing so might flag that as an apnea. In this case, the flow graph might show an arousal before the apnea rather than after.

OSCAR shows what the CPAP machine reports, so there should not be a discrepancy between OSCAR and the machine.

It is always possible that OSCAR is not displaying the flag in exactly the right spot in the flow graph. This can happen when the CPAP machine reports event times with a different resolution than it uses for the flow graph. I haven't looked at the DeVilbiss loader in some months so don't recall how closely the event flags displayed should match the flow graph.

There's about a fifteen second delay, i.e. actual events usually precede the flags on Oscar by about fifteen seconds. But I'm used to that and know where to look.

It can't hurt to get an oximeter if you can afford one. I'd recommend getting one that's compatible with Oscar:

OSCAR is currently compatible with Contec CMS50D+, CMS50E, CMS50F and CMS50I serial oximeters.
(Note: Direct importing from bluetooth models is probably not possible yet)
You may wish to note, other companies, such as Pulox, simply rebadge Contec CMS50's under new names, such as the Pulox PO-200, PO-300, PO-400. These should also work.

This, and more, can be found by clicking on the Oximeter Import Wizard link on the Welcome page in Oscar.

Miss Emerita wrote: ↑

Sat Nov 27, 2021 12:59 pm

It can't hurt to get an oximeter if you can afford one. I'd recommend getting one that's compatible with Oscar:

OSCAR is currently compatible with Contec CMS50D+, CMS50E, CMS50F and CMS50I serial oximeters.
(Note: Direct importing from bluetooth models is probably not possible yet)
You may wish to note, other companies, such as Pulox, simply rebadge Contec CMS50's under new names, such as the Pulox PO-200, PO-300, PO-400. These should also work.

This, and more, can be found by clicking on the Oximeter Import Wizard link on the Welcome page in Oscar.

My original sleep study showed my average Ox Sat as 87% and it dropped as low as 63% at one stage. So I guess if I can get something which is capable of linking with a phone app and/or storing a night's data, it may be enough. Must do some research.

MMcG wrote: ↑

Sat Nov 27, 2021 8:36 am

There's about a fifteen second delay, i.e. actual events usually precede the flags on Oscar by about fifteen seconds. But I'm used to that and know where to look.

I've just reviewed the code and the positioning of the flags in the chart. The CPAP machine seems to be providing a timestamp for the event that is "near" the event in the flow graph, but not exactly correct. The timestamp is usually after the event, but by how much is variable. I'll see if I can do anything to get the flag closer to the event. I doubt I can improve it much.

About oximeters, OSCAR also supports the Viatom/Wellue line.