Ten seconds of calibration on other shafts gave no clear decrease in later SEEG high-gamma power error in 36 patients

We tested a fixed high-gamma power forecast in 36 patients with two public interictal SEEG records each. Ten seconds in file A selected eight bipolar pairs. A shared correction used ten seconds in A and B from shafts with no selected pair. The test used B seconds 90 to less than 100. Mean absolute error changed from 2.302 to 2.268 dB. Corrected minus unchanged error was -0.033 dB, with a 95% paired patient interval of [-0.101, +0.033] dB. This test gave no clear mean decrease. Direct calibration of selected pairs gave 1.082 dB error. Its advantage over the other-shaft rule was 1.186 dB, with interval [+0.591, +1.907] dB. We preserve all features, forecasts, code, source records, and four figures. Independent arithmetic, fresh source checks, and exact repeat runs passed. The file calendar order is unknown. This secondary analysis measures recorded interictal power, not task decoding or stimulation effects.

stereoelectroencephalography · interictal power · high gamma · calibration · brain-computer interface · forecast error