a difference between SH & Oscar

[sheepless]

all you math whizzes out there may have already dismissed the issue I raised above regarding the disparity between max values in SH and Oscar. I'm a little slower. my math skills have always been pretty weak, and weaker with age it seems, but upon further reflection IDK if there's a problem there or not but am slowly realizing, I think, that the difference between 99% and 100% isn't necessarily 1%, which is what I initially based my assumption on, that there might be an additional max calculation problem. feel free to correct me. I suspect this is pretty elementary arithmetic and I'm a little embarrassed (disconcerted even) by my uncertainty. lingering brain fog, I hope.

[Crimson Nape]

Have you tried exporting the daily data in question and preform the percentile function on it in Excel?  99 Percentile and 99 Percent are totally 2 different animals.

Example:
11 numbers in order of input: 8, 4, 11, 15, 7, 6, 5, 8, 8, 3, and 1
Sum = 76
Average = 6.909091
50 Percent = 38
99 Percent = 75.24

- Compared to: -

50 Percentile or Median = 7
90 Percentile = 11
99 Percentile = 14.6

To obtain a percentile, you first organize your data set by order of magnitude.
This would make " 8, 4, 11, 15, 7, 6, 5, 8, 8, 3, 1" to be listed as "1, 3, 4, 5, 6, 7, 8, 8, 8, 11, 15".
Next you look at (or extrapolate) the value at the position's relative position.  Example: The value of 7 in the Median, or middle, of the data set and corresponds to the 50 Percentile value. Using the 90% from the beginning of the data returns the value "11" in that location.

[Gideon]

a 95% value is not 95% of the max value but the value at which 95% of the data points are below.
Max is and should always be Max.  It was changed in SHv1.1.0 to remove outliers.  In reality max anything means very little since it is usually an outlier.  We will find another way of handling that.

[sheepless]

so max figures are percentiles? I didn't see that explicitly noted in the program but makes more sense given the 95% context as bonjour notes. which suggests that I was initially mistaken (and the difference between 100% and 99% is not 1%) and the figures I noted above may not be miscalculated. sorry for the confusion and thanks for the explanations.

[Crimson Nape]

so max figures are percentiles?

No, a max value is the absolute highest value recorded.  Using flow rates as an example, if you took you  mask off while your CPAP was still running, it would record this flow rate as your max value, but it has nothing to do with your actual therapeutic flow rates.  In an effort to remove these spikes, or artifacts, is why SH and OSCAR started using the 99 percentile value as the max value.


Edit: I just reread your post and realized that I read it incorrectly.  You are correct! the max value listed in SH 1.1.0 and OSCAR is the 99 percentile value and NOT the highest recorded.
Sorry!

Side note: when attempting to use the emoticon for "sorry", which use to be the word sorry enclosed by colons on each end, produces this weird symbol.

Code:

:sorry:

results in: orry:

Oh Well . . .

[sheepless]

sorry, I meant the max value that's not a max but said to be 99%. plus I was thinking true max and 100% can be expressed as the 100th percentile even if doing so doesn't tell us much more than max or 100%. but I'm really out of my element here as much schooling has faded.

[Crimson Nape]

You are CORRECT Sir! I was editing my post above yours with a correction and got side tracked.

[GuyScharf]

Guy, does SH 1.1.0 do the same?  SH1.0.0?

To summarize the behavior after further testing, SH 1.1.0 and OSCAR ResMed importer:

1. If a day has one or more sessions with detailed data, any summary sessions are completely ignored.
   
2. If there are only summary sessions, then all summary sessions for the day are reported.  There is no statistics data.  Sample day attached.
   

   

[19cisne36]

This may not be the case, but as a new user I found very soon that ResMed thinks, that no one can sleep over midday or noon.
As a night worker, going to bed at 8 am and waking up 4 pm shoved me, that I had slept only 4 hours at that night...
I solved the problem (with help of this forum, thanks!) adjusting the ResMed clock 7 hours behind. So if I'm not sleeping at 7 pm (which is only once in a blue moon), stats are correct for me.

But do not know, if this is anything related to the topic...

[sheepless]

a variant of this issue: yesterday I downloaded from the sd card. my last session was short and I wasn't asleep so I turned it off. I don't think this has anything to do with it but mention just in case. returned card to the machine. napped before noon. today I downloaded from sd card. in SH 1.0.0 beta 2 broken gl, yesterday's nap session and nap data appear in yesterday's SH data. Oscar ignores the nap. no session listed and no data. there are no summary sessions in either program, unless a turned off session is treated like a summary session. I did not close the program yesterday or today (some misbehaviors seem to show up only after closing and reopening the program).

the difference between this incidence and the one at the top of this thread is: the first time the data wasn't recorded on the sd card because it was locked; this time the data is on the card but Oscar ignored it.