Eight fixed median gamma SEEG readers had 0.500 mean worst contact Brier error under five simulated 1 mV pulses across five segments at 10 percent active time

Eight fixed median gamma readers were tested on public SEEG windows from OpenNeuro ds004473 version 1.0.2. Five simulated 1 mV contact faults had 50 active samples in a 500-sample window. The faults had equal pulse count, peak voltage, active time, and raw energy. Mean worst contact Brier error was 0.225637 when all faults stayed in one power segment. It was 0.499947 when the faults reached all five segments. The paired increase was 0.274310, with a descriptive 95 percent person bootstrap interval of [0.259206, 0.299236]. All eight changes were positive. Balanced accuracy at the selected contacts fell from 63.47 percent to 50 percent. Three pulses in three segments also gave error near 0.500 at 6 percent active time. The result shows that short active time does not ensure an unchanged majority of segment powers. The package contains exact windows, fixed models, saved scores, source ranges, checks, code, and figures. This is an exploratory test on reused data with synthetic voltage faults. No stimulation or live interface was tested.

stereoelectroencephalography · gamma power · median spectral estimation · synthetic artifacts · Brier error