I would also be very curious if this was happening to me.
Here are some ideas of what I might try ...
Set up some sort of external monitor for a couple of nights and see what happens while I was asleep.
A phone camera (plugged in to a power source) running in time lapse mode could do it,
or an audio recording app left on overnight might give some hints.
Also I would noodle over what else is observable in the data that might give more hints:
.. Are there any discernible patterns in the data gaps,
same time each night,
same amount of elapsed time since you turned the machine on each night
number of gaps per night,
time intervals between gaps,
are gaps the same duration,
similar data patterns in any metrics leading up to the gaps,
number of nights in between occurrences,
etc. ?
.. Does the number of hours reported on the machine display match the total sleep time in the data or are the gaps not counted?
.. When you look at the data in Sleepyhead, does the summary report look different from the detail report?
.. Do you have an alternate software program to review the data and compare it to Sleepyhead reports?
Do they report the same gaps as each other?
Maybe try a new SD card to eliminate possibility of dead spots on the media?
This is an interesting puzzle. Hope you solve it. And let us know the answer
Saldus Miegas