CSV discrepancies

[DanEm]

Hello everybody.

Has some of you know, I work a lot with graphs and to do so, I use Oscar to export data to work in Excel.

I came upon something I cannot reconcile and I ask for your input.

Consider the results for February 13.

I had only one session of sleep from 2023-02-13T23:23:08 to 2023-02-14T07:14:10.
On Oscar Daily view, 95% Flow Limit. is shown as 0.05
In the corresponding CSV file (exported session), 95% Flow Limit. is shown as 0.18
That is more than 3 times the value in the Daily view.
The gap (error) is 260%.  
Since Oscar shows FL with the precision of 1/100 (2 digits), the above is out my comprehension.

The included screenshot shows some of the verification I did.

Am I missing something?  How can this be explained or rectified?

[DanEm]

I have rechecked everything and still found discrepancies.

The AHI values from the CSV files match the AHI values on the Daily View, but the FL values do not.

The following graph shows the results of my research.

   

I really want to know which set of values is the right one.

Please, comment this  

[pholynyk]

Hi Dan - A similar problem was reported earlier by JasonR - I poked t it a bit and found:

>>>"It's not just the Flow Limitations, the Pressure and EPAP values also differ between the Statisitics Panel and the csv file.
>>>
>>>Clearly, more investigation is needed, and I will report back as I find out more."


I haven't been able to investigate further, and I haven't figured out which is correct, if any.

It has occurred to me that we could write down the machine displayed values and compare them with the statistics values and the csv values... the issue here is that I'm not sure if the machine is displaying the median or some othe percentile value. I haven't looked in the Clinicians guide to se it it says which.

That's one bit of checking you could do... The machine doesn't display flow limitations, but we could compare the displayed values ... I believe the Statistics panel of the Daily page will be correct, but I have no evidence yet.

I did do a brief code reading, and didn't notice any discrepancies, so it needs more work.

[pholynyk]

Ahh, here we go; from the AS10 Clinicians manual:

Code:

Pressure Average pressure during the selected period (95 th percentile for each day; average of
the 95 th percentile values for periods >1 day).

Leak Average of the 95 th percentile values of leak during the selected period for days with
usage only.

AHI Apnea-Hypopnea Index—average AHI during the selected period. AHI and AI are
calculated for times of low leak only.

[DanEm]

Hi polynyk.

Thank you very much for coming back on this issue.
The big question is which FL values are the good ones.

Pressure and EPAP discrepancies have no impact on therapy nor the results.  Not knowing the precise values does not impact the results.

In my case, the average discrepancy between CSV and the values shown on the Daily View is 0.1. It is bothering since 0.1 is a target I was given (I think it was Gideon) and my results can go above or under depending on the real value.
My FL average with the CSV file data is 0.23
My FL average with the Daily Vue data is 0.14

These values are not corrected for data to exclude because of testing I did.

I have the "Resmed-AirCurve-10-Vauto-S_clinical-guide" and I went through it without seeing anything about any of this.  I will go at it again.

I am not certain how far we can check what the AirCurve is displaying.  To my knowledge, it shows only the current day values. It is only a one-point check but I will do it.

I also found that if you uncheck some sessions in the Daily View bottom left panel, the Statistics reflect the change (depending on what is unchecked, some values in Statistics go down - AHI is one of them).

Just saw your second post.  I suppose it describes the values shown on the AirCurve screen.

[pholynyk]

The AirCurve machines have some additional items:

Code:

Pressure Average inspiratory pressure during the selected period (95 th percentile for each day;
average of 95 th percentile values for periods >1 day).
Exp. Pressure Average expiratory pressure during the selected period (95 th percentile for each day;
average of 95 th percentile values for periods >1 day).
Leak Average of the 95 th percentile values of leak during the selected period for days with
usage only.
Vt Average of the 50th percentile values of tidal volume during the selected period for
days with usage only.
RR Frequency of breathing, expressed as the number of breaths per minute (5-breath
moving average).
MV Average of the 50th percentile values of minute ventilation during the selected period
for days with usage only.
Ti Duration of inspiration (ie, the respiratory flow into the lungs), expressed in seconds
(5-breath moving average).
I:E I:E is the ratio of the inspiratory period to the expiratory period.
Spont Trig Percentage of breaths that are spontaneously triggered, measured from the last 20
breaths.
Spont Cyc Percentage of breaths that are spontaneously cycled, measured from the last 20
breaths.
AHI Apnea-Hypopnea Index—average AHI during the selected period. AHI and AI are
calculated for times of low leak only.
Total AI Apnea Index—average total AI during the selected period.
Central AI Central Apnea Index—average CAI of the Days Used in the selected period.

I see that flow limitations are not displayed, but if we can match up any of the displayed values with either the Statistics panel or the csv report, I think we can assume that the fl are the same.

[DanEm]

I'm quitting for the day, will pick it up tomorrow.

I checked in the AirCurve report and only one set of data is shown.
It is a 30 day backward span.

My AHI is shown as 0.4.
Oscar Statistic shows 0.44 for the same backward span.

If it is proof that Oscar Statistics are the same as the AirCurve, then if I want my FL data I will have to move one day at the time and note each FL value in Statistics.
A bit of a pain and the thing would probably be multiplied by the number of sessions if I want all those too.

Have a good night 

[DanEm]

I have concerns about my therapy, but that question is expressed in the main forum.

I am wondering if the Details CSV export would be helpful to get some answers here.
It is a pain to work with because of the size but if something can be gained I would not mind doing it.

[pholynyk]

Dan , I'm guessing you can't see all those other items on your display. I think that means you need to go into the Cliniciians menu and change Essentials from 'On' to 'Plus'. You then should be able to change the reporting period from 30 days to 1 day.

Then you can compare the displayed values to the values in the Statistics panel of the Daily page and the values in the csv file.

This will make it easier for me to know which chunks of code to study.

[DanEm]

Ghee, just when I am ready to send my compare results

My setting for Essentials is already on Plus.

What I did not know was that I could modify the 30 days reporting 

I will make the change and go over it again.

Just in case something can be saved, take a look at my Excel screenshot.