(Another) Respironics Encore Data Failure

Doric wrote

I'm sure you've read here that any changes should be made one at a time and give each change a few nights to settle in. I learned that the hard way at the beginning when I made too many changes at one time and couldn't figure out what was working and what was not.

For pressure and CPAP modality changes, I agree 100% as these are adaptive changes, ones your body and mind, even subconsciously, must adapt to before you can assess the results (+ or -) of the change. For many other changes (only one per sleep period) that I am making and analyzing (both scientifically and statistically), I don't agree. If something shows promise, I try it again (and again), usually on a different night and for a different sleep period. If the results from the change continue to be positive, I incorporate it into my sleep therapy. I am pleased with my process so far (9 weeks and counting) as I have progressed from a weekly AHI of over 8 with over 2% Time in Apnea to last week's result of a 0.7 AHI with a 0.06% Time in Apnea (1.8 minutes total out of 50.2 therapy hours). My wife has also received sleep benefits as my snoring is now non-existant (used to keep her awake). I have already reached a level of restful sleep that, three months ago, I would have believed impossible to achieve. Again, I can't thank the folks on this forum enough for providing through their postings the education, knowledge, recommendations, and tips that I needed to accomplish this satisfying and beneficial result.

Cinco777 wrote

Re improved granularity/quantification, the Encore Pro Analyzer provides accurate counts of the sleep events and Apnea time per sleep interval - Encore Pro reports/views do not. Encore Pro's graphical view (Sleep Therapy Daily Details, Page 4) shows sleep events as tick marks and does not use a Time-Of-Day (TOD) axis. Early on, when my AHI was > 5, I found it impossible to ascertain if a tick mark represented one, or more than one, event. I also found it time consuming to associate a sleep event (tick mark) to the TOD that it occurred when I wanted to try to match it to an external event such as my switching sleep positions. When I have time, I plan to use the Encore Pro SQL event tables or parse the exported XML file and create a time-stamped sleep event TOD timeline that provides much finer event granularity.

Using the Encore Pro SQL database tables and the associated exported XML file as inputs, I have created a Microsoft Excel (2003) spreadsheet that is allowing me to determine if any (strong/quantifiable/predictive/probable, i.e., statistically reliable) correlations exist among my sleep events - Snore, Flow Limitation, Hipopnea, Obstructive Apnea, and Non-Responsive Apnea, and the reported leak values. This time-stamped source data is collected by my Respironics M-Series Auto CPAP w/Aflex and recorded to SmartCards. I have marked each time-stamped event with a code: D=Leak(13), E=Snore(14), F=Flow Limitation(16), H=Hipopnea(17), O=Obstructive Apnea(18), and R=Non-Responsive Apnea(19). Leak events for my Respironics unit also include the Leak Value (Liters/minute) and "leak end" date and time (the date and time when the leak rate changes to another leak rate, either higher or lower). I am currently not evaluating Variable Breathing(15). I sort all the time-stamped events (after splitting the date (mm/dd/yy), and time (hh:mm:ss) into separate columns so I can visually see :ss). The end result is a time-sequenced table of my events with the event code as the first column of my spreadsheet. For analyses purposes, I calculate the delta time from the previous event and the duration @ a specific leak value (leak rate). My Respironics unit only reports leak values in multiples of 7, i.e., 18, 25, 32, 39, ... LPM - I don't know if this holds true for other manufacturers/units. I am using Excel functions plus some Excel Add-ins to do my statistical analyses.

Cinco777 wrote

When I have time, I plan to use the Encore Pro SQL event tables or parse the exported XML file and create a time-stamped sleep event TOD timeline that provides much finer event granularity.

I created a MS Excel (2003) template for charting my nightly sleep events. The chart includes leak, snore, flow limitation, hypopnea, obstructive apnea, and non-responsive apnea (I've only experienced two and that was in early April). I've also created another template that adds Variable Breathing. I use the tables in the Encore Pro SQL database as input. I have color coded the events so they are visible on my display (1600x1200) and on printouts (Epson Inkjet). I am posting two sample charts. My next goal is to automate the process so I can create a Night Event chart in a few minutes each morning. A long term goal is to provide a "Zoom" capability when viewing the charts. My Variable Breathing %, as reported by the Encore Pro Analyzer, has averaged less than 10%/night the past month, with one night registering 1.5%. I plan to look at VB to see if it can tell me something, good or bad, about my sleep apnea therapy.

And again. Despite being aware of my previous "double click" experience, and being as careful as I can - my dexterity just doesn't appear to be up to the task (expecially at 4:00am) - I did it again. Very odd, though, this time. After a fairly slow double click on the center button, the cool blue lights went out and the blower stopped. About a second later, the lights came back on, and the blower started again. It seemed similar to the rebooting of a computer. One more press to turn the blower off.

This morning, of course, I found that my four hours of data prior to 4:00am were nowehere to be found. Everything after that was fine.

And this morning I tried to replicate the experience. No success. Varied the timing between the two presses, varied the pressure on the switch. In each case, the machine simply responded as it should to the two presses. Perhaps it's designed to fail only at night?

This is my third time losing data for the starting hours of the night, and my second time recognizing that a "double click" of the center button was involved. Never lost data ever for the latter part of a night, or for an entire night.

As an unrelated surprise, I've also found that the blower will SOMETIMES turn itself on when there is suction on the hose. (REMstar has an auto-off option, but I can find nothing similar to ResMed's "SmartStart" option.) This also is very hard to replicate; amount of suction, speed of suction, duration of suction - nothing makes it happen consistently. Just turns on when it wants to, it seems. Or perhaps I have gremlins?

I lost all data for Sunday night. We had a blip in the power in the wee hours of the morning due to thunderstorm. Not sure if it is related or not. I probably had not been on the machine 4 hours when it occurred but for sure was on it more than 4 hours after the power came back on. Saturday results are non existent. I don't think that I touched any of the buttons (but can't swear to it) and autostart kicked in immediately when I put the mask back on after removing it when the momentary power outage occurred. Encore Viewer not Encore Pro.

Velbor wrote:As an unrelated surprise, I've also found that the blower will SOMETIMES turn itself on when there is suction on the hose. (REMstar has an auto-off option, but I can find nothing similar to ResMed's "SmartStart" option.) This also is very hard to replicate; amount of suction, speed of suction, duration of suction - nothing makes it happen consistently. Just turns on when it wants to, it seems. Or perhaps I have gremlins?

Sounds like the "Auto-ON" feature which is affected by the mask leak rate and seal plus hose length, HH, etc.

Velbor wrote:And again. Despite being aware of my previous "double click" experience, and being as careful as I can - my dexterity just doesn't appear to be up to the task (expecially at 4:00am) - I did it again. Very odd, though, this time. After a fairly slow double click on the center button, the cool blue lights went out and the blower stopped. About a second later, the lights came back on, and the blower started again. It seemed similar to the rebooting of a computer. One more press to turn the blower off.

Sounds to me like the power cable could have been moved or loose? Since you couldn't duplicate the trouble by pushing the button?

GumbyCT wrote:Sounds like the "Auto-ON" feature which is affected by the mask leak rate and seal plus hose length, HH, etc.

Aha! Yes, thank you. Just found this in the "Glossary" section of the clinician manual and the user manual. Always enabled, unlike ResMed where it can be turned on or off. No reference to it anywhere else in either manual that I can find. Never ran into this before; I always turn the blower on before breathing. Odd that it was so difficult for me to get it going by sucking on the tube (no lewd responses, please).

Bad day for replication (and again re: lewd responses).

Pugsy wrote:I lost all data for Sunday night. We had a blip in the power in the wee hours of the morning due to thunderstorm. Not sure if it is related or not. I probably had not been on the machine 4 hours when it occurred but for sure was on it more than 4 hours after the power came back on. Saturday results are non existent. I don't think that I touched any of the buttons (but can't swear to it) and autostart kicked in immediately when I put the mask back on after removing it when the momentary power outage occurred. Encore Viewer not Encore Pro.

Pugsy, That seems to be the "famous" glitch in the M Series. I and many others have contacted Respironics. They're aware of it and "are working on it". It seems to happen on the 6-7th day, usually the weekend, when it seems to "run out of room". I'm not too savvy about this but thanks to Hawthorne, I use this utility program every Sun AM to format it and have had no problems since about Oct, except for last weekend when I lost data again. I think it's because my hubby was sleeping 9-10hrs for most of the week(he's OK, rainy weather I guess) and the card again "ran out of room" It's frustrating and I think, outrageous. Dori

http://www.mycpap.org/

Velbor,

You may remember that with my nightly "multiple smartcard procedure" and saving/formatting my SCs each morning, I have never experienced a data loss. You and others have and I still see postings where members report a data loss. Since it appears at night, infrequently, and may be related to multiple off/ons of the blower occurring in a short time period (a few seconds), I would like to propose a theory for when & possibly why these data losses are occurring.

My theory is that the data loss is occurring when the blower is turned off/on repeatedly in a very short time interval about the same time that the Respironics unit switches to its new Noon-to-Noon day.

The available evidence/suspicions seem to support my theory. For US time zones, this always occurs at night. For West Coasters like me, the machine's Noon-to-Noon day now ends at 5AM PDT (UTC/GMT time - 7 hours). For East Coasters, the machine's day ends at 8AM. Multiple users are reporting data loss without clear knowledge of when it happened (but usually recognized after waking and downloading/looking at their night's sleep data). Even with your skill level, you have lost data and not been able to recreate a data loss. I believe that you would be successful if you do your experiment(s) around the Time-of-Day (TOD) that your machine switches to its new Day. I do not know what Time Zone you reside in so I do not know if you would have to run your experiment(s) at 5AM like me or at a more reasonable TOD such as 8AM (if you live on the East Coast).

Since Respironics has not fixed this problem, I suspect that it is inherent in their Noon-to-Noon design which uses UTC/GMT for their internal clock. It may be a hardware implementation but is, more likely, in the firmware. Too bad that the US Government can't (won't) do a recall since CPAP is a medical device only available via a RX.

If no one else investigates this in the coming month, I may put in on my To Do list as I am my usual curious self.

cinco777 wrote: My theory is that the data loss is occurring when the blower is turned off/on repeatedly in a very short time interval about the same time that the Respironics unit switches to its new Noon-to-Noon day.

Review of my records of partial-night data loss shows:

18 May 09: loss of data prior to 4:46am local time (EDT)
31 May 09: loss of data prior to 1:54am
15 Jun 09: loss of data prior to 4:04am
25 Jul 09: loss of data prior to 4:11am

I've never lost data at a UTC-significant time, and I've never lost an entire day.
And the losses I have had are not particularly consistent!
Sorry that my data isn't more cooperative with your theory! Regards, Velbor

Velbor wrote

I've never lost data at a UTC-significant time, and I've never lost an entire day.
And the losses I have had are not particularly consistent!
Sorry that my data isn't more cooperative with your theory! Regards, Velbor

My proposed theory doesn't require that the actual lost data be "near" to or span the TOD that the machine changes from one day to its next, only that something (multiple presses of the on/off button, power outage (even a momentary one), etc., occur when the machine is preparing to switch from its current Noon-to-Noon day to its next day. As a result of this "interruption" near its switchover time, the machine (likely firmware) reinitializes certain pointers it uses for writing to the SmartCard (similar to a table of contents or directory) and then overwrites, at least partially, previously recorded data (or, at least, corrupts the previously written data so that EncorePro and EncoreViewer discard it as invalid). I wish the author of the MyCPAP card utility would take an interest in this problem. Since he "knows" the format of the data written on the SmartCard, he could, if provided card images of the SmartCards involved in the data loss, see if any of the standard data structures are corrupted. Without his help, we have to recreate the data loss and discern the cause(s) and effect(s).

Thanks for the additional observational data re your own data losses. I believe that if everyone had observational records as good as yours for when data losses occur and which data is lost, we would have a much better understanding of this data loss problem and, possibly, know why they occur.

It looks like come October, I will be awake at 5AM (my machine's switchover time) trying a few experiments to, hopefully, lose some data. Wish me luck??

I experienced a partial data loss last night and posted about it here: viewtopic/t47248/viewtopic.php?p=426128#p426128

I wrote:I slept for nearly ten hours last night, which felt good. But I lost a good chunk of data this morning. I went to bed just before 10:00 p.m., got up to use the bathroom at about 4:30, and then continued sleeping until 7:45.

When I went to shut the machine off this morning, I think my finger bounced on the on/off button just enough to re-start the machine immediately, because the pressure kept coming, and I had to hit the on/off button again to stop it. When I looked at my data, it ends at 4:30, and there's no record of my usage between 4:30 and 7:45.

After re-reading this thread, I find it interesting that Velbor's partial loss of nightly data (due to the same "extra button push" that I was guilty of this morning) resulted in his losing the earliest portion of data for that session; in my case, the most recent portion of data is missing.

Just thought I'd throw my experience into the mix of this interesting thread. More fodder for you statistics junkies.

I stepped on my surge protector not 1, not 2, but 3 mornings in a row, and lost all data for those 3 days. I wonder if the "double on/off hit" acts at a "power off"?

While running and saving my weekly reports this morning, I noticed a little sliver of "usage" at the tail end of my 12/4 data (close to 8:00 a.m.), which I'm guessing was recorded when I mistakenly "bounced" the on-off button. I zoomed in on the chart to see it better, and it clearly shows the gap in data recording:



Rather than the recorded sleep time of 6:41 for that date, it should have recorded close to 10 hours. I goofed by pushing the button more than once, and the machine goofed by losing more than three hours of data. I hate it when that happens.