Figure 5. Fluctuations in the ultra-slow modulation of the path causes changes in the dynamics of the seizure.

(A) A recording in which the seizure begins to go toward a SupH offset (square-root decreasing amplitude), but the amplitude increases again and the final offset is SH. (B) A portion of the simulation of the model in Saggio et al., 2017 done to reproduce the dynamics observed in the recording. This timeseries is high pass filtered to simulate the effect of AC recordings. The non-filtered simulation is shown in (C). (D) A portion of a long status epilepticus recorded in one patient. The status epilepticus was characterized by transitions in the dynamics (such as at 70 s and 230 s). (E) High-pass filtered, and (F) unfiltered portion of model data simulating the behavior in D. (G) A zoomed, flattened projection of the map in Figure 4B (see Appendix II, IV), for reference in H-I. (H,I) Amplitude and frequency maps showing trajectories (white) of the modelled seizures in C,F. The seizure from C begins (red triangle) and naturally moves upwards toward the supH bifurcation, but a change in the ultraslow drift prior to termination pushes the system downward, changing the offset from SupH to SH. Conversely, the seizure from F begins (black triangle) and repeatedly moves towards the SH termination due to the ultraslow drift, but is pushed back into the seizure regime repeatedly by high levels of noise, which overrides the role of the slow variable in terminating the seizure.