Is the S9 event data questionable at best?

[brucifer]

For the past week I have been faithfully downloading data from my S9 SD card into the ResMed and SleepyHead software. Yesterday at noon I had planned to take a nap. I had just put on my mask, had my S9 cranked up, and had just gotten into the bed when my daughter sent me a text. I was hoping our text conversation would be short, so I left the mask and equipment on while we exchanged text messages. Ninety minutes later our texting ended, and I decided to skip the nap, so I turned off the machine and thought nothing else of it.

Today when I downloaded my data and analyzed the wave graphs and event break downs, I noticed that during the time period that I was awake and texting that the S9 recorded data that was very similar to data recorded when I was sleeping. For example, during the 90 minutes of awake time, five OAs were recorded, three CAs, and one H. Those events were all false positives. However, the frequency of these data pretty much align with my averages for the week. My AHI for the week was 4.05. I find it rather peculiar that I had events recorded while I was awake that matched my events in frequency of occurrence while sleeping. The data comparison makes me wonder the true efficacy, accuracy, and usefulness of the event data from the S9. Anyone care to comment on this?

[BlackSpinner]

For the past week I have been faithfully downloading data from my S9 SD card into the ResMed and SleepyHead software. Yesterday at noon I had planned to take a nap. I had just put on my mask, had my S9 cranked up, and had just gotten into the bed when my daughter sent me a text. I was hoping our text conversation would be short, so I left the mask and equipment on while we exchanged text messages. Ninety minutes later our texting ended, and I decided to skip the nap, so I turned off the machine and thought nothing else of it.

Today when I downloaded my data and analyzed the wave graphs and event break downs, I noticed that during the time period that I was awake and texting that the S9 recorded data that was very similar to data recorded when I was sleeping. For example, during the 90 minutes of awake time, five OAs were recorded, three CAs, and one H. Those events were all false positives. However, the frequency of these data pretty much align with my averages for the week. My AHI for the week was 4.05. I find it rather peculiar that I had events recorded while I was awake that matched my events in frequency of occurrence while sleeping. The data comparison makes me wonder the true efficacy, accuracy, and usefulness of the event data from the S9. Anyone care to comment on this?

Most people's breathing while awake is very erratic. We hold our breath when we move or when we get startled. We do deep breathing when excited. When we sleep we don't do that, breathing goes on auto pilot as we sleep.

[sol]

The machine is fooled, cannot tell if you,re asleep or wake. Even if you,re asleep there is no way of knowing if its a true apnea or not, an apnea have to be accompanied with oxygen desat

[pacificpap]

Interesting. I tried to take a nap yesterday but couldn't fall asleep (good sign, I hope), and I was surprised today to see that the S9 logged four obstructive apneas, three hypopneas, and two CA events within a 15 minute period during which I'm sure I was awake and pretty sure I wasn't breathing erratically. Zooming in on the flow rate in Sleepyhead, it looks different than event clusters that happen when I am asleep and the flow rate drops to almost nothing.

[brucifer]

Most people's breathing while awake is very erratic. We hold our breath when we move or when we get startled. We do deep breathing when excited. When we sleep we don't do that, breathing goes on auto pilot as we sleep.

True. However, I was on my back and very still while texting away with my iPhone. I was breathing very normally except for the Tidal Volume and the Minute Ventilation. Those two wave charts showed much greater amplitude. The point I am making is that my S9 was unable to distinguish between apnea events that occur while I am sleeping and false-positive events that occur while I am awake. Consequently, if my S9 cannot distinguish between what is a real event and what is not, then what is the point of depending on the data generated? In essence, we really don't know what we are looking at when we view our respective data.

[brucifer]

The machine is fooled, cannot tell if you,re asleep or wake. Even if you,re asleep there is no way of knowing if its a true apnea or not, an apnea have to be accompanied with oxygen desat

Sol, that is interesting. I'm not familiar with oxygen desat.

pacificpap, I didn't really have any clusters, but I did know that the frequency of OAs, CAs, and Hs were pretty much the same when I was awake and when I was asleep. MY AHI was also very similar during both period types.

[pacificpap]

Yeah, I have clusters Even moderate leaks for me result in huge AHI spikes, and I'm pretty sure those are legit.

Looked again, and it seems like a combination of lower flow rate overall (shallow breathing?) and intermittent leaks caused the false positive events from my non-nap yesterday. In contrast, the flow rate during my nighttime OA events shows a almost a flat line preceding the event markers. Just a guess, but it looks like the events are identified via pattern recognition applied to the flow rate, and maybe there's an amplitude threshold for that which can be triggered by shallow breathing. Just a WAG from a software developer.

[brucifer]

Yeah, I have clusters Even moderate leaks for me result in huge AHI spikes, and I'm pretty sure those are legit.

Looked again, and it seems like a combination of lower flow rate overall (shallow breathing?) and intermittent leaks caused the false positive events from my non-nap yesterday. In contrast, the flow rate during my nighttime OA events shows a almost a flat line preceding the event markers. Just a guess, but it looks like the events are identified via pattern recognition applied to the flow rate, and maybe there's an amplitude threshold for that which can be triggered by shallow breathing. Just a WAG from a software developer.

Yes, I see the same thing with my graphs. My flow rate recorded during my non-nap had less overall amplitude than my flow rate during sleep, yet the spikes during sleep triggered apnea-related event clusters. I also noticed that leaks during sleep trigger event clusters as well. That being said, it is difficult to determine what is real and what is a false positive. At the end of the day, we can only guess.

[Pugsy]

I would rather have a machine that might score a few false positives instead of a machine that missed scoring real positives.

[BlackSpinner]

Most people's breathing while awake is very erratic. We hold our breath when we move or when we get startled. We do deep breathing when excited. When we sleep we don't do that, breathing goes on auto pilot as we sleep.

True. However, I was on my back and very still while texting away with my iPhone. I was breathing very normally except for the Tidal Volume and the Minute Ventilation. Those two wave charts showed much greater amplitude. The point I am making is that my S9 was unable to distinguish between apnea events that occur while I am sleeping and false-positive events that occur while I am awake. Consequently, if my S9 cannot distinguish between what is a real event and what is not, then what is the point of depending on the data generated? In essence, we really don't know what we are looking at when we view our respective data.

No your were not breathing normally for a sleeping person. There is a very big difference between a sleeping person's breathing and one who is awake. A sleeping person breathes much deeper and slower then a person who is awake and concentrating. Awake breathing with concentration tends to be very shallow and it would feel like an event to the algorithm.

[brucifer]

I would rather have a machine that might score a few false positives instead of a machine that missed scoring real positives.

And how do you tell the difference between the two? That is my point with this thread.

[pacificpap]

I wonder if during the sleep study the EEG data is used as the primary diagnostic for events as well as sleep stages. Or if it's correlated with the PAP/O2 data. I also recall sensors being placed on my upper lip to measure airflow. It seems like the machines' algorithms would be tested for accuracy using those other metrics by the manufacturers, too.

Perhaps I have OWA along with OSA

[archangle]

The machine is fooled, cannot tell if you,re asleep or wake. Even if you,re asleep there is no way of knowing if its a true apnea or not, an apnea have to be accompanied with oxygen desat

It's also considered to be an apnea if you have an "EEG Arousal", which indicates your sleep was disturbed.

CPAP machines can only look at airflow and pressure.

Apneas don't "count" if you're not asleep. CPAP machines can't tell if you're awake. You may even be "awake" during the night, but not remember it in the morning.

If you stop breathing, but don't do it long enough to disturb your sleep or cause O2 to drop, it's not considered to be an apnea and probably isn't harmful. CPAP can't see this either.

I think it's important to look at the airflow waveform and see how long and severe the "apnea" is. If it's 15 seconds a few times an hour, no big deal. If it's 120 seconds of zero airflow, it's probably a big deal even without checking O2 or EEG.

CPAP will probably show some false apneas. However, it will tend to show all the "real" apneas, and if you DO have bad apnea, it will tend to show bad apnea.

[archangle]

I wonder if during the sleep study the EEG data is used as the primary diagnostic for events as well as sleep stages. Or if it's correlated with the PAP/O2 data. I also recall sensors being placed on my upper lip to measure airflow.

If you put a CPAP mask on a patient, you have to run a minimum pressure in order to flush the exhaled air out of the mask and hose. The airflow sensors on the nose are mostly so they can measure airflow at zero pressure with no mask. i.e. it measures your apnea without treatment.

[robysue]

The machine is fooled, cannot tell if you,re asleep or wake. Even if you,re asleep there is no way of knowing if its a true apnea or not, an apnea have to be accompanied with oxygen desat

On a PSG, an OA or CA does NOT require an O2 desat. It does require that you are asleep (according to the EEG) when the OA or CA occurs,