After Mikesus showed how to access the Encore Pro database I set about writing an Excel spreadsheet that would generate additional plots (hopefully useful ) that are not available in Encore Pro itself.
It's still a "work-in-progress". Here are three pdf files showing the output as of today:
Sheet 1 contains four plots based on all the data available: a) the number of hours at each pressure for the whole database. (note that Encore Pro rounds-up pressures to the next integer value so you don't see 8.5 cm for example, it's included as 9 cm), b) fraction of time spent below each pressure, and c) and d) the calculated total AHI computed at each pressure. These two plots have different data overlaid. On the lower left plot, the pink squares represent the nightly values of AHI plotted against the REMStar Auto's minimum pressure setting. The blue dotted line represents the average of these points. On the lower right plot the pink squares represent the nightly AHI plotted against the average APAP pressure for that night.
Click here for Long Term Pressure Summary data
Sheets 2 and 3 both contain the same long term trend data, but are divided into the months of February and March 2005.
Click here for February trend data
Click here for March trend data
My question is - can anybody think of any additional plots/charts that might be useful to us?
derek
Extra plots derived from Encore Pro Database...
And here are the same plots for another person (who may want to remain anonymous since I didn't get his permission to post them )
Click here for Long Term Pressure Summary data
Click here for February trend data
Click here for March trend data
These plots are not as complete because I did not have all of the necessary data available.
derek
Click here for Long Term Pressure Summary data
Click here for February trend data
Click here for March trend data
These plots are not as complete because I did not have all of the necessary data available.
derek
-
littlefeat
- Posts: 36
- Joined: Sun Mar 20, 2005 10:29 am
Derek and Mikesus--
Thanks again to you guys for working to make more robust data usage possible and available from the Encore Pro/Remstar Auto setup. A few questions/comments/suggestions:
1. Would these same data show with respect to CPAP or C-Flex modes, or only Auto mode? Are they now showing with respect to Auto w/ C-Flex, or simple Auto only?
2. I think that your sheets 2 and 3 represent basically the same data that comes in the canned Encore reports with respect to sleep therapy trends, but can come from multiple downloads instead of only single downloads. Correct?
3. I know that someone said that Leak is in a separate table in the database, but Leak info on longterm, multi-download sleep therapy trend reports (like sheets 2 and 3) would be really helpful, as leak interacts with pressure and the efficacy of delivery of the pressure necessary to address As and Hs.
4. I need to study more the two bottom plots on sheet 1 to see what they really are conveying, but I'm sure that they will be useful.
5. Would it be difficult to show, on an individual night view, the "real" pressures/snores/FLs/As/Hs/leak minute-by-minute, taking into account C-Flex pressure releases and ramps, and show the pressures un-rounded? One of the troublesome things about the canned views is that the database presumably is capturing minute-by-minute (or more frequent) data, but there's no way to "drill down" into any more detail than a nightly, smoothed graphic representation. This goes to your prior comment that the database reporting is overly normalized or averaged or rounded, even though presumably the individual, precise data values are in there.
Best regards--
Thanks again to you guys for working to make more robust data usage possible and available from the Encore Pro/Remstar Auto setup. A few questions/comments/suggestions:
1. Would these same data show with respect to CPAP or C-Flex modes, or only Auto mode? Are they now showing with respect to Auto w/ C-Flex, or simple Auto only?
2. I think that your sheets 2 and 3 represent basically the same data that comes in the canned Encore reports with respect to sleep therapy trends, but can come from multiple downloads instead of only single downloads. Correct?
3. I know that someone said that Leak is in a separate table in the database, but Leak info on longterm, multi-download sleep therapy trend reports (like sheets 2 and 3) would be really helpful, as leak interacts with pressure and the efficacy of delivery of the pressure necessary to address As and Hs.
4. I need to study more the two bottom plots on sheet 1 to see what they really are conveying, but I'm sure that they will be useful.
5. Would it be difficult to show, on an individual night view, the "real" pressures/snores/FLs/As/Hs/leak minute-by-minute, taking into account C-Flex pressure releases and ramps, and show the pressures un-rounded? One of the troublesome things about the canned views is that the database presumably is capturing minute-by-minute (or more frequent) data, but there's no way to "drill down" into any more detail than a nightly, smoothed graphic representation. This goes to your prior comment that the database reporting is overly normalized or averaged or rounded, even though presumably the individual, precise data values are in there.
Best regards--
Fat Man in the Bathtub
The problem is that the data is scattered across numerous tables. Yes each event down to the second is in the DB, however, to tie it to anything meaningful is daunting. The Database itself is overnormalized, so data from each type of machine is in their own tables, and data appears to be included more than once in multiple tables. What I would really like to do is find a way to grab the data before it hits the database. At this time tho, I don't know if that is possible.
Whoops - I got Mike's plots messed up - now corrected.
Interesting how his AHI/pressure curve is similar in shape to mine. They seem to indicate that the AHI is reduced down to an almost constant residual level, but not eliminated, with increasing pressure.
I think that the data points at the low and high ends are probably unreliable because of the very small time spent at those pressures.
Makes me wonder if there is any point to treatment above 10 cm H2O for Mike and me? There doesn't seem to be any benefit in terms of improved AHI.
Interesting how his AHI/pressure curve is similar in shape to mine. They seem to indicate that the AHI is reduced down to an almost constant residual level, but not eliminated, with increasing pressure.
I think that the data points at the low and high ends are probably unreliable because of the very small time spent at those pressures.
Makes me wonder if there is any point to treatment above 10 cm H2O for Mike and me? There doesn't seem to be any benefit in terms of improved AHI.
Derek,derek wrote:Whoops - I got Mike's plots messed up - now corrected.
Interesting how his AHI/pressure curve is similar in shape to mine. They seem to indicate that the AHI is reduced down to an almost constant residual level, but not eliminated, with increasing pressure.
I think that the data points at the low and high ends are probably unreliable because of the very small time spent at those pressures.
Makes me wonder if there is any point to treatment above 10 cm H2O for Mike and me? There doesn't seem to be any benefit in terms of improved AHI.
Thought your pressure didn't go that high? Curious why the graph wouldn't show a flat leveling as you suggested... To me the graphs actually look identical... Very odd...
What I think would be a great graph would be a line graph with AHI Pressure, Leak, AI, HI, FL, and OA, and NRA. This would give us a good picture of why the pressure was raised. I am not sure if the data in histogram is detailed enough for it tho. The respective events tables have the data, but as you noticed before it appears to repeat if you do not erase data...
Littlefeat -
There are (to my count) 83 tables in the database. Most are useless. The one that Mike and I have looked at the most is dbo_Histogram, which seems to be a summary of events. For each type of event, however, there is a table that has the occurence of every single thing that happened. The problem is that it only has the start time recorded (at 30 second steps). there are no other details such as the duration of an apnea.
I've searched around but have been unable to find any reference to C-Flex.
There are (to my count) 83 tables in the database. Most are useless. The one that Mike and I have looked at the most is dbo_Histogram, which seems to be a summary of events. For each type of event, however, there is a table that has the occurence of every single thing that happened. The problem is that it only has the start time recorded (at 30 second steps). there are no other details such as the duration of an apnea.
I've searched around but have been unable to find any reference to C-Flex.
The only reference to Cflex is the description in unitmode table. It does not relate to any data in the table as far as I can tell...derek wrote:Littlefeat -
There are (to my count) 83 tables in the database. Most are useless. The one that Mike and I have looked at the most is dbo_Histogram, which seems to be a summary of events. For each type of event, however, there is a table that has the occurence of every single thing that happened. The problem is that it only has the start time recorded (at 30 second steps). there are no other details such as the duration of an apnea.
I've searched around but have been unable to find any reference to C-Flex.
-
Guest
Sounds like dbo_SleepTrendEventLog might be helpful to pull plots from. The line graph proposed by Mikesus a couple of posts ago sounds great because it would capture leak as well.
Respironics should have someone as smart as you guys working on development of this thing. Sounds like there is a lot of sound and fury (signifying nothing) in what could have been a very elegant and pwerful database tool.
Any thoughts on the Auto vs. CPAP mode questions that I posed earlier?
Thanks again for your efforts, Mikesus and Derek.
Best--
Respironics should have someone as smart as you guys working on development of this thing. Sounds like there is a lot of sound and fury (signifying nothing) in what could have been a very elegant and pwerful database tool.
Any thoughts on the Auto vs. CPAP mode questions that I posed earlier?
Thanks again for your efforts, Mikesus and Derek.
Best--
-
littlefeat
- Posts: 36
- Joined: Sun Mar 20, 2005 10:29 am
Sounds like dbo_SleepTrendEventLog might be helpful to pull plots from. The line graph proposed by Mikesus a couple of posts ago sounds great because it would capture leak as well.
Respironics should have someone as smart as you guys working on development of this thing. Sounds like there is a lot of sound and fury (signifying nothing) in what could have been a very elegant and pwerful database tool.
Any thoughts on the Auto vs. CPAP mode questions that I posed earlier?
Thanks again for your efforts, Mikesus and Derek.
Best--
Respironics should have someone as smart as you guys working on development of this thing. Sounds like there is a lot of sound and fury (signifying nothing) in what could have been a very elegant and pwerful database tool.
Any thoughts on the Auto vs. CPAP mode questions that I posed earlier?
Thanks again for your efforts, Mikesus and Derek.
Best--
Fat Man in the Bathtub
Mikesus wrote:The only reference to Cflex is the description in unitmode table. It does not relate to any data in the table as far as I can tell...derek wrote:Littlefeat -
There are (to my count) 83 tables in the database. Most are useless. The one that Mike and I have looked at the most is dbo_Histogram, which seems to be a summary of events. For each type of event, however, there is a table that has the occurence of every single thing that happened. The problem is that it only has the start time recorded (at 30 second steps). there are no other details such as the duration of an apnea.
I've searched around but have been unable to find any reference to C-Flex.
-
littlefeat
- Posts: 36
- Joined: Sun Mar 20, 2005 10:29 am
AS Mike suggested I made some extra plots showing the dependence of some other indices on pressure.
Click here for derek's pressure dependence
My distribution of pressure graph has changed because I found a "buglet" in my spreadsheet.
Click here for Mikesus's pressure dependence
Note that the values for Mike at 8 cm and 16 cm will be very unreliable (large potential errors) because of the very small time spent at those pressures.
Mike I suspect you have duplicate sessions in your data - would that be correct? It should not affect the trend too much.
Click here for derek's pressure dependence
My distribution of pressure graph has changed because I found a "buglet" in my spreadsheet.
Click here for Mikesus's pressure dependence
Note that the values for Mike at 8 cm and 16 cm will be very unreliable (large potential errors) because of the very small time spent at those pressures.
Mike I suspect you have duplicate sessions in your data - would that be correct? It should not affect the trend too much.