Parse overnight data possible?

[BlueSky1958]

Had a night where the dog got sick so I had to jump out of bed abruptly and shut off machine. Restarted machine 10 min. later, but my last 2 hours were restless and much worse than prior 6 hours of normal undisturbed sleep. Is there a way to delete the bad data and preserve/keep the normal data?

[Crimson Nape]

Deselect those time periods in the lower left of the Daily screen.

- Red

[BlueSky1958]

Right, but won't that "bad"data re-import each time I import in the future? Can it be removed permanently?

[Crimson Nape]

Nope. Otherwise a person could alter the data to invalidate the true reporting. At that point, no one would trust OSCAR as being accurate.
- Red

[BlueSky1958]

It may be my naivete as a new CPAP user, but what reason or benefit would there be to anyone altering their data? Is it required by some insurance carriers? My doctor hasn't mentioned it or asked for it (I donlt know if shes even aware of what OSCAR is). My only motivation is to remove "abnormal" outlier data to get a more accurate trend over time. This is done in research all the time. I anticipate this would not be a very common occurrence, but nonetheless, when I look at last night, my last 2 hour segment completely pollutes the prior 6 hours and also my trend. If it was part of an uninterrupted sleep session during which there was no abnormal interruption , I wouldn't touch it because then it would be valid.

[Crimson Nape]

Here is the first thing that comes to my (feeble) mind. While your intentions are good, someone could remove good data to make it look like they need a more expensive model.

- Red

[OpalRose]

.......If it was part of an uninterrupted sleep session during which there was no abnormal interruption , I wouldn't touch it because then it would be valid.

Many of us experience "interruptions" during the night.  Weather we need to get up to use the restroom, get a drink, we simply can't sleep or our pets/children wake us, etc., does not mean that that these are abnormal interruptions.  It's just part of life.

If you normally sleep straight through the night without interruptions, you are probably in the minority.  

With that said, I do understand what you're saying.  Those extra sleep sessions could show a lot of apnea or just be the opposite....

I had 4 sleep sessions last night. I have been sick for a week now with strep throat, and not able to sleep. I think you get to a point, when these wake ups and "extra" sleep sessions won't bug you so much, but I do understand.

The OSCAR team has worked and continue to work hard to offer a free software package that is comprehensive and as accurate as possible.  OSCAR only reports what your Cpap is reporting.  

It's not meant to be altered by the user.  Just take the suggestion above and turn off the second session for that particular night, so you won't see it.

[BlueSky1958]

Hmm, I guess, although it feels like the built-in limitation is designed around the few bad actors at the expense of helping the majority of good users with non-nefarious intentions.

Are you referring to a person who uses their OSCAR data to manipulate their healthcare provider into ordering the more expensive equipment? Does OSCAR data and output hold that much sway and authority? I'm unaware so give me some rope here.

Two ideas:

1) As open source, there must be a way to save the data file (with outliers excised) as a separate file that you can append good data to over time. However, the raw data still exists on the SD card and that cannot be changed, so the provider would and should always ask for that?

OR....

2) Allow users the option to "flag" data segments they feel are not representative of normal sleep (such as my 2 hours last night). The user could toggle a radio button/check box which would effectively ignore (but not delete) any and all previously flagged segments, and therefore their trend/monthly/cumulative stats would show a more representative output/report/summary and trend. If someone or entity requires the "raw data view," build in a watermark or some sort of way to validate an unaltered report. All the data on the card itself always stays unaltered.

[BlueSky1958]

Your comments lead me to believe my second idea (presented in last post) may be a really good option and compromise to consider. It gives users flexibility without locking them into having to accept data outlier segments they don't feel are representative. I think flexibility is almost always better (that's why I choose Android over Apple   ).

If some users want to include every interruption over the course of a night even if that amounts to 6 separate segments of restless sleep, they have that option. Others may take the approach that uncommon interruptions during sleep, while it happens, is still an outlier and sits outside their normal trend (and should not be included). Throwing out outliers is common practice in all research. Your goal isn't to change the trend because you want it to look better, as much as it is to get a more "pure" view of trend that more accurately represents your experience.

Example: If I have 26 nights out of 30 where my AHI is below 3, but on four nights I experience an interruption that causes me to lose just and only the last 3 hours of sleep (tossing, turning, thinking about stuff) and my AHI one those nights goes to 12, all because 12 crappy hours were counted for that month, is my AHI for the month representative of the other 228 hours I logged at <3? I would say no, just as a cholesterol test taken 4 hours after you eat a bucket of fried chicken with a milkshake will not be representative of your normal cholesterol score had you fasted 12 hours prior.

OR I could just leave my machine off the rest of the night....lol. But...

In my suggestion, unaltered data stays that way on the card. There is no change. No modification. They only change is the ability for the user to toggle on/off to exclude all previously flagged "bad" segments. You could also lock out the option to save a file or report i f the "toggled" option is selected. I'm not sure how you'd effectively do that with the current tool  once you accumulate many days/months worth of data. Too cumbersome and time consuming I suspect.

Just food for thought. Love OSCAR.

[Crimson Nape]

The presence of this "bad" data, that you wish to remove, can act as an alert to other problems with your sleep session.   Its removal is kind of like tearing down your garage, so no one can see how messy you keep it.

- Red