A Hann gamma envelope feature did not lower mean Brier error beyond matched power on later trials in eight reused SEEG records

We tested whether gamma envelope concentration adds useful information after a Hann taper reduces the weight of buffer edges. We used 500-sample windows from eight reused SEEG records in OpenNeuro ds004473 version 1.0.2. The reader distinguished rest from a state before a known key response. A common baseline had 40 power features, including gamma power from both candidate spectra. Adding eight Hann envelope features changed mean later-trial Brier error from 0.166819 to 0.167577. The paired change was +0.000757. Its 95% person bootstrap interval was -0.000921 to +0.002589. Error fell in three of eight people. The interval permits small gains and losses. In 4096 synthetic signals with random phases before tapering, mean phase-corrected concentration changed from -0.006234 to +1.408843 after the Hann taper. Thus the phase correction does not remove the processing effect of the taper. We reproduced the earlier comparisons, checked all 88 final models with a second optimizer, and reproduced all 24 analysis output files exactly. Fresh EDF ranges matched the selected test voltages in all records. A full source rebuild matched all 128 trials in the smallest record. The package preserves exact voltages, provenance, source retrieval code, features, fits, predictions, three figures, and repeat instructions. This is an exploratory result on reused data with fixed inherited selection. It does not establish live BCI performance or a neural cause.

stereoelectroencephalography · signal processing · gamma activity · Brier score · brain-computer interface