False Positives for OA and H

[roster]

No offense taken. I just did not want to leave the impression that I support "chasing" a small number of apneas and constantly tweaking the pressure.

I take a strong position, in and out of the forum, that individuals, if they so choose, should have commercially available tools to improve their therapy beyond what one night in a sleep lab can accomplish. But obsessing over the data is counterproductive.

"Perfectionists pursue the perfect life and many go to the grave having missed the good life."

[brain_cloud]

I've noticed my data recorded some OA and H during a time I know I was awake. Knowing this, doesn't it call into question the validity of the data which records events when I'm sleeping? How do I know if the OA's and H's actually occurred or are just another false positive? Last night I scored an AHI of 2.0, but 3 of the 7 OA's I know for a fact were false positives in the first 15 minutes of CPAP.

One thing that I found useful was to have a short test session on the machine (using a watch with a second hand) where I was attempting to induce mchine scored "events" to get a sense for what it responds to. One of the things I found suprising (and I have a different machine) was how quickly the machine would ramp from min to max pressure (12.5 to 15.5) upon a few light "feigned" snores. And I realized how easily natural variation in my breathing (awake, during the test) could get scored as hypopneas or very short (10-11 second) apneas. Just things like moderate to deep sighs and whatnot. I'm pretty sure that I will always have a residual level of machine-scored hypopneas and short apneas that are not really a problem for me. Around 3 per hour on average. I'm kind of looking for clusters of apneas and when those are absent, I'm satisfied, even though my AHI might be 5 or even higher.

In the interests of full disclosure, I am still a newbie with just 10 CPAP nights to my name.

[Silver Pelt]

Newbie or not, you are using the machine data to your advantage. I harp on not chasing numbers because some folks agonize over scores like yours.

Welcome to the club.

[williamco]

Yes, I had apnea while I was fully awake. it was training time for my mask and machine in the middle of the day for an hour or so and I had AHI of 4

it is not like it was at beginning of sleep and I don't remember when I was awake and when I wasn't. it was just an isolated hour in middle of the day, so the laps of memory is not there

[cinco777]

Like Jason S., quoted below, I suspected, when I started CPAP in March '09, that some of my Apneas and Hypopneas occurred when I was awake and were, therefore, False, and I didn't have to pay any attention to them. For the first two months of my CPAP-enabled therapy, I split my nightly As & Hs into two columns in my spreadsheet: real events and false events (when I thought I was awake). After two months of trying to convince myself that what I was doing was valid and useful in improving my therapy, I gave up - I didn't have the data and couldn't acquire the data (EEG) to make an accurate determination of when I was sleeping and when I was awake.

Velbor's view of this subject, which I just read today, having been directed here by a current posting to another thread, describes what I concluded from my observations and failed analyses. Beginning in June '09, I treated all my As & Hs as real and worked to eliminate them. I succeeded in reaching my AHI goal of < 1 and am now focusing on sleeping through the night.

Jason S. wrote:
I've noticed my data recorded some OA and H during a time I know I was awake. Knowing this, doesn't it call into question the validity of the data which records events when I'm sleeping? How do I know if the OA's and H's actually occurred or are just another false positive? Last night I scored an AHI of 2.0, but 3 of the 7 OA's I know for a fact were false positives in the first 15 minutes of CPAP.

by Velbor on Tue Sep 01, 2009 4:35 pm

Many important issues here.

First, breathing while awake can be very irregular. It's no big deal to hold our breaths, either intentionally or unknowingly, for 10 seconds or more, while awake. While moving. While daydreaming. There is no a priori reason to doubt the machine sensors. The machine doesn't know our state or depth of sleep. It just records and analyzes and reports pressure and flow patterns.

There is nothing in this which calls into question the "validity" of data recorded while sleeping.

As to "false positive" results, the question is, false positive WHAT? If we expect that the numbers churned out by the machinery reflect ONLY events DURING SLEEP, then indeed we are setting ourselves up to be misinformed. We have to acknowledge that the numbers reflect events for the ENTIRE PERIOD that the blower is on, which includes everything from awakeness to light sleep to deep sleep to REM sleep, each present multiple times through the night, and likely different each night. There's something not quite right in selectively tossing out "awake" data at the beginning of the night, when there is likely more "awake" data as the night progresses that we're not aware of and don't remember in the morning.

It's our expectations that need to be clarified! The scientist faced with anomalous data questions their own assumptions first. The possibility of lab equipment error must always be considered, but that's usually rather far down the list.

Good observation. Velbor

[roster]

Like Jason S., quoted below, I suspected, when I started CPAP in March '09, that some of my Apneas and Hypopneas occurred when I was awake and were, therefore, False, and I didn't have to pay any attention to them. .......... I didn't have the data and couldn't acquire the data (EEG) to make an accurate determination of when I was sleeping and when I was awake.

....... Beginning in June '09, I treated all my As & Hs as real and worked to eliminate them. I succeeded in reaching my AHI goal of < 1 and am now focusing on sleeping through the night.

I went through a similar progression as Cinco and agree with Velbor that the possibility of machine error in recording events is rather far down on our list of concerns.

Thanks for the post Cinco.

[flylow]

As already stated here, these machines determine obstructive and hyponeas based on limited data as opposed to all of the sensors used in a sleep study. So when the sleep lab can titrate to an ahi of zero with a leaky mask, and you can not come close to that at the more comfortable environment of home, there must be false positives. It is inevitable. This does not mean that the data is useless. A trend is a trend and less false positives likely means a lower index. Folks do warn of placing too much importance on the data since it is inherently flawed but I will happily always chase the data in pursuit of the perfect night and, more importantly, consistent therapy.

Ultimately you do need to evaluate how you feel versus what the data tells you. My only caveat to this is when you get consistently bad therapy, you do tend to acclimate to however you feel and may not recognize the functional deficits you are living through. I was in denial before being diagnosed with severe OA.

These discussions always leave me unsatisfied as there doesn't appear to be any detection algorithms published. I think it would be enlightening but this information is probably held as a trade secret.

[roster]

These discussions always leave me unsatisfied as there doesn't appear to be any detection algorithms published. I think it would be enlightening but this information is probably held as a trade secret.

Apnea Detection - The Respironics REMstar algorithm establishes a baseline of patient flow based on a moving flow signal window - it detects an apnea as an 80% reduction in flow lasting at least 10 seconds.

Hypopnea Detection - The Respironics REMstar algorithm establishes a baseline of patient flow based on a moving flow signal window - it detects a hypopnea as a 40% reduction in flow lasting at least 10 seconds, followed by a recovery breath.

Snore Detection - Snoring is a measure of partial airway occlusion and is a strong indicator of potential respiratory events. The Respironics REMstar algorithm has the capability to detect snoring vibrations and adjust the pressure accordingly.

Flow Limitation - The Respironics REMstar algorithm analyzes the flatness, roundness, peak and shape of each breath throughout the night and determines a gross change in these characteristics as a flow limitation.

Non-Responsive Apnea / Hypopnea - At higher pressures, if a patient does not respond to pressure increases due to apneas or hypopneas, the device will automatically convert to a CPAP for a 15-minute period and an NRAH notification will be conveyed to Encore.

http://encore.respironics.com/type.aspx

[flylow]

These discussions always leave me unsatisfied as there doesn't appear to be any detection algorithms published. I think it would be enlightening but this information is probably held as a trade secret.

Apnea Detection - The Respironics REMstar algorithm establishes a baseline of patient flow based on a moving flow signal window - it detects an apnea as an 80% reduction in flow lasting at least 10 seconds.

Hypopnea Detection - The Respironics REMstar algorithm establishes a baseline of patient flow based on a moving flow signal window - it detects a hypopnea as a 40% reduction in flow lasting at least 10 seconds, followed by a recovery breath.

Snore Detection - Snoring is a measure of partial airway occlusion and is a strong indicator of potential respiratory events. The Respironics REMstar algorithm has the capability to detect snoring vibrations and adjust the pressure accordingly.

Flow Limitation - The Respironics REMstar algorithm analyzes the flatness, roundness, peak and shape of each breath throughout the night and determines a gross change in these characteristics as a flow limitation.

Non-Responsive Apnea / Hypopnea - At higher pressures, if a patient does not respond to pressure increases due to apneas or hypopneas, the device will automatically convert to a CPAP for a 15-minute period and an NRAH notification will be conveyed to Encore.

http://encore.respironics.com/type.aspx

Thanks for that, but this information is way too general to understand the nuances that are inevitable. Useless drivel to most, but as an engineer, I like to dive deeper.

[peppi]

Jason:
I have confirmed what you suspected in the first few nights . Waking hours really can not be considered in the HI and AI numbers. My breathing isn't rythmic when I'm awake. Unfortuntely there is no way to strip that number out of the total AHI reported on the machines display.
This is the main reason I bought the USB adapter and software to monitor exactly when the events are occuring so I can manually discount those events.

I know I'm getting excellent sleep because I feel wonderful so most of this stuff is moot but I just like knowing. (I'm an engineer too)
Get the software and look further into it if it bothers you too.

Best,
Phil

[roster]

..........................
To guard a title that was rich before,
To gild refined gold, to paint the lily,
To throw a perfume on the violet,
To smooth the ice, or add another hue
Unto the rainbow, or with taper-light
To seek the beauteous eye of heaven to garnish,
........................

-WS

[cinco777]

Thanks for that, but this information is way too general to understand the nuances that are inevitable. Useless drivel to most, but as an engineer, I like to dive deeper.

Search this forum for Respironics/Resmed/... patents. Some members of this forum (SWS, Muffy, and a few others) have analyzed these patents and frequently offer guidance and explanations on understanding the algorithmic and design details of our machines and how our machines operate in addressing Sleep Disturbed Breathing (SDB).

I've printed, read, and reread all the patents that I found links and pointers to, and am still learning new, interesting, and helpful things about my machine and what makes it tick. I highlight insightful sections when I read and some of the Respironics (my machine brand) patents now look like a rainbow.