Ok, that _is_ weird.
Firstly, I'm glad that I'm not the only one who has issues with their SleepyHead plots and axis labeling. With your mask pressure as labeled, you are getting up to 35 cmH2O pressure at the mask, which is significantly above the capabilities of your machine.
One thing I did notice was that your breathing rate is unusually slow in those plots - something like ~10 breaths per minute. I wonder if what is being shown here is a sequence whereby you are solidly in Central Apnea (ie. you are not breathing on your own) and the machine is essentially acting as a ventilator. Hence the increased flow rate when the pressure is increased as the air is pushed into your lungs, and the sharply negative flow rate when the pressure is reduced as your resting state pushes the air (rapidly!) out of your lungs.
The SleepyHead software is not fully compatible (yet) with PR BiPAP autoSV Advanced machines. The weird scales on the vertical axes may be due to a bug that affects those machines. JediMark (aka Mark Watkins), the author of SleepyHead has said that he does not yet have enough data from ASV machines to fully debug the SleepyHead problems that affect the PR BiPAP ASV users who try to use SH. If you're willing to share your data, you might want to register over there and contact JediMark through a PM.