I'm just starting out into the world of sleep apnoea and though I don't (yet) use a CPAP, I bought an oximeter (CMS50E ver 6.6R) and somewhere along the line, I found out about SleepyHead.
So far so fairly good, but I've found a few minor issues and the SourceForge SH mailing lists (both -dev and -user) appear to be invitation-only (or, at least, requiring admin/mod approval), the -user list seems abandoned and so I'm not quite sure where the best place to report them is, how to help fix them or even whether SH is still an actively maintained project considering the last publicly available build appears to be rather over a year old (though I can see that Mark is still making occasional commits to the git repo).
ETA: SH Help » About says:
Quote:SleepyHead v0.9.8-1 (testing)
Build Date: Oct 3 2014 17:17:35
Revision: 4595153 (master branch)
which is why I believe(d) the software to be over a year old, even though I installed from SleepyHead-Testing-0.9.8-1-T3-Windows-Installer.exe, dated 28 October, 2015. I'm now confused as to whether the software I'm running is 2 or 14 months old.
Before I file any bug reports at SF, I will get the Qt SDK and MingW tool chain and try building (Win 7, 64-bit) the current git master/HEAD in case any of these issues have already been fixed. (That will be fun, because most of my dev experience has been under Linux.)
Meanwhile my question is, are there any devs other than Mark involved and, if so, is there any place in lieu of the usual SF mailing lists where I can talk to them and coördinate our activities?
In brief, those issues I've discovered are:
- Importing by serial download seems to result in very noisy data (bad samples, eg SpO2 of 0, 40 or 128% (magic number there, probably indicating lost bytes in the serial stream) and 0 PR), where uploading the same recording via .spoR file works as it should. Where the bad data is located varies from attempt to attempt. I can provide an example of a .spoR that imported correctly where direct download didn't, and an archive of a profile showing the bad imports.
- Imported data, whether by serial d/l, live recording or from .spoR, will end up recorded against the day prior to the date specified on the calendar in the import dialogue if the hour of the recording is before noon of the selected day (I assume this is related to the day split time in preferences). This can happen if you went to bed after midnight, or you keep strange hours like I do (it's getting on for 6am here now!).
- Changing the oximetry thresholds etc doesn't refresh events, even though the source code seems to suggest that it should (but that may be a reflection of the age of the build). The event list can be forced to refresh by running Data » Automatic Oximetry Cleanup.
What else that operation does, I don't know, because I've not been able to find any documentation on it (nor on the effect of the option 'Discard segments under $x seconds' in Preferences » Oximetry).
My other issue, which is not a fault of SH, is that it does not appear to be possible to selectively delete or edit data once imported (other than to delete a whole day's worth of data, which is okay if you have a .spoR recording for that day).
The reason I wanted to do that was because, being new to both the oximeter and SH, of course I experimented under different conditions (waking vs sleeping, recorded on the pulse-ox vs recorded live sessions with plethysmographs etc) with the result that there are some very abnormal readings (including an SpO2 of 44% when I moved it from one hand to the other!), consequently the stats are nonsense and also the SpO2 plot won't autoscale properly (because it was using the 44% as the minimum). The autoscaling problem is a minor nuisance but easily worked around by manually setting the y-axis range, but it looks like the stats problem requires getting rid of the bad data.
Evidently, the thing to do is to maintain two profiles, one for sleep-only data and the other for experimentation. I discovered issue #2 after I created a new profile to import my night-long .spoR files.
However, those recordings also contain waking data, and I need to trim those samples out or the stats will be wrong for the sleep-only profile. Looking into the way SH stores data, it seems that it's a lot easier to edit the .spoR files than it is to edit the SH records directly, so I've written a Python script to do just that. I can send it to anybody who wants it, and maybe it's even something worth adding to the SH repo. (It currently doesn't support .spo2R files, but that wouldn't be hard to add.)
Sorry for the long-winded first post, but I hope the folks here can point me in the right direction for how best to liaise with Mark and the other devs, if any.