Which AHI Numbers Should I Believe, from Sleepyhead or ResScan?

[OpineCone]

My sleep doctor decided a month ago to make a big reduction in my APAP pressures (long story). He told me to watch the results, especially the AHI, and to increase the pressures a bit if necessary to keep the AHI below 5.0.

I seemed to be doing well, based on the data I downloaded every six days to ResScan. During the period from 3/2/17 through last night, the largest AHI reading was 2.4. At least, that's what I thought.

Today I became interested in examining my results in Sleepyhead, with an expectation that I would see essentially the same data but with more details or more informative charts. To my surprise I discovered that my AHI numbers in Sleepyhead were completely different from those in ResScan. Most notably, the largest AHI reading recorded in Sleepyhead was 4.94, for the same night for which ResScan reported the AHI as 2.4 (the night of 3/4/17). See below for the relevant charts from ResScan (top) and Sleepyhead (bottom).

This raises some interesting and perhaps disturbing questions. How can the same data show very different statistics in the two programs? Which AHI should I believe? Am I doing something incorrectly? And so on.

I always transfer my data from the SD card first to ResScan and immediately afterwards to Sleepyhead.

Some help with this situation will be much appreciated.

[OpineCone]

It occurs to me that this thread might belong in the Software Support Forum. Perhaps a moderator would kindly consider whether to move it.

Also, I suppose it might be easier to answer my questions if I provide more detailed information. My ResScan statistics for 3/4/2017 show an AHI of 2.4, while my Sleepyhead chart for the same day shows an AHI of 4.94. The relevant ResScan table and Sleepyhead chart are shown below. I am wondering why there is a discrepancy. There do not seem to be any discrepancies between my ResScan and Sleepyhead AHI numbers during the two-year period prior to 3/2/17, which is the day my sleep doctor told me to lower my pressure settings from 17-11 to 11-4, but there have been continuing discrepancies since the pressures were changed.

Does anyone have any suggestions as to what has gone wrong, please?


[attachment=3340]


[attachment=3341]

[PaulaO2]

Take a look at the bottom section of the left sidebar of your SleepyHead screen. See those doubled sessions? That is the problem. Click the green button next to the time of one of the first two. Take a look at the other graphs and make sure they are still there. If they are, you turned off the correct one. If, for example, it was the second of the 23:59, then click the green button for each second of the doubled entries.

This happens to me a lot. Usually it happens for days I had already downloaded but for some reason, it downloads it again, adding the doubled sessions.

2.4 doubled is 4.8 but if you add in the different ways the two software views the data, it's about right with your result of 4.94.

[Sleeprider]

Good catch Paula. I have never seen this and will watch for it in the future on other's data. Could this be a S9 thing?

[OpineCone]

Thanks to PaulaO2 for sharing her information about the doubled sessions and resulting overstated AHI numbers. I confirm that turning off one session of each doubled session does indeed correct the AHI number shown on the Sleepyhead Daily screen.

I believe there are several other related issues that will now need attention, including: why the doubled sessions appear, how we can stop them, how we can correct our historical data, who is affected by the bug and who isn't, whether there is a need to notify people of the bug, and whether the software developer should be consulted.

In an attempt to determine the severity of the bug, I counted the number of nights for which my Daily screen shows doubled sessions during the last few months: zero in November and December 2016, one in January 2017, fifteen in February 2017 (2/4, 2/6, 2/7, 2/8, 2/9, 2/10, 2/12, 2/18, 2/19, 2/20, 2/21, 2/24, 2/25, 2/26, 2/28), and ten in March 2017 (3/2, 3/4, 3/5, 3/6, 3/8, 3/12, 3/13, 3/14, 3/18, 3/21). I also went back two years and found zero double sessions in February 2015.

Two conclusions:

1. When it happens, it happens a lot.
2. It probably did not happen on my system prior to 2/4/17. Why it started then, I do not know.

The tedious counting revealed another aspect of the bug: the doubled sessions are never identical. Some combination of the start time, the end time and the duration differ by a small amount, usually two or three seconds.

Also, there is another kind of anomaly, observed for the night of 3/11/17, in which one session is 4h20m longer than the other but both end at almost the same time, just one second apart.

Correcting historical data seems to be possible, by finding doubled sessions and turning off one of the two. The corresponding changes do not show in the Overview chart until you close and reopen Sleepyhead. I assume the Statistics screen is also updated but I have not checked.

To date, the only people to have reported this bug are users of ResMed S9 AutoSet machines. It would be helpful if users of other machines would check their Sleepyhead history for double sessions and other anomalies. The best method seems to be to click through the Daily screens and look at the CPAP Sessions data in the Session Information for each day. Any night showing more than one session requires examination of start times, end times and durations. I know from experience that it can get quite complicated when you had three or more machine sessions during a night and some of them are doubled. If you find an easier way, please let me know.

[Sleeprider]

Opine, one easy solution may be to rebuild the CPAP data. Go to the menu, Data, and rebuild the data. This will either solve the problem or not, but won't hurt anything.

[OpineCone]

Thank you, Sleeprider. Rebuilding the data resolved all the doubled sessions and corrected the corresponding AHI numbers, except the weird anomaly for the night of 3/11/17 is still present. Sleepyhead shows two overlapping sessions that night, which is of course impossible. One session starts at 23:48:17 and ends at 08:11:33, while the other runs from 05:08:32 until 08:11:32. ResScan does not list sessions but the ResScan chart shows there were two separate sessions that night with a break of approximately 20 minutes between sessions.

Is there another tweak you would suggest to resolve the Sleepyhead anomaly for that night? The Sleepyhead and ResScan charts are shown below.

I think it may be significant that I recently upgraded to the current version of Sleepyhead, on March 2 2017. I would guess that the next time I imported data from the SD card after the upgrade, Sleepyhead created the fifteen doubled sessions in February and two or three at the beginning of March. It is unclear to me exactly why there were more doubled sessions later in March, but that doesn't seem to matter now.

[Sleeprider]

The other tweak is to purge the data for that date from Sleepyhead. It should re-import the day next time you put in your SD card. It's just one day, so not a critical loss if you don't get it back. The option is Data/Advanced/Purge current day. Be sure you are looking at the day in question.

[OpineCone]

The other tweak is to purge the data for that date from Sleepyhead.  It should re-import the day next time you put in your SD card.  It's just one day, so not a critical loss if you don't get it back.  The option is Data/Advanced/Purge current day.  Be sure you are looking at the day in question.

I followed these directions, including verifying that the data for 3/11/17 was purged by observing the "Nothing's Here!" screen and then following the normal process of importing data from the SD card. However, the new Sleepyhead screen for that night is almost exactly the same as it was before the purge and re-import. It still shows two overlapping sessions, even though that is impossible. The only difference is that the start of the first session has changed from 23:48:17 to 23:48:19.

In order to verify the validity of the data on the SD card, I closed SleepyHead, opened ResScan, deleted the 3/11/17 sessions from ResScan, saved the file, and then followed the normal process of importing data from the SD card to ResScan. The ResScan chart for 3/11/17 has been restored and looks identical to the previous chart.

Conclusion: The SD card data is fine, will import into ResScan, but will not import correctly into SleepyHead. Worse, it takes close examination of the SleepyHead charts to determine that a single day is displayed incorrectly. I probably would not have noticed the issue if the doubled sessions had not also occurred.

Is it time to get the SleepyHead software developer involved? Might it help if I were to make a copy of my SleepyHeadData-Testing folder (3.09 GB) or some portion of it? And how about a copy of my SD card (42 MB)?

[Sleeprider]

The Sleepyhead developer is unlikely to try to repair this isolated bug. Mark Watson is health-challenged and we are hoping that the software will keep up with new machine issues, let alone improve on existing issues. I think this is something you can live with. The point of using the software is to evaluate your efficacy trends. Freeware doesn't promise perfection. If you have ResScan, then that is the official software. If it shows a similar error, then the problem is the data written to the SD card.