Gamma/beta selection gave fewer seizure onset labels than gamma selection for eight SEEG pairs in 36 epilepsy patients

We compared two rules that select eight adjacent SEEG pairs from 60 seconds in 36 epilepsy patients. One rule ranks 70–110 Hz power. The other ranks that power divided by 13–30 Hz power. The selected seizure onset label fractions were 21.2% and 4.5%. The mean paired change was -16.7 percentage points. Its 95% patient bootstrap interval was -24.3 to -9.7 points. The ratio retained less gamma power in 35 of 36 patients. Mean power relative to the best eight test-record pairs fell from 83.5% to 20.0%. The planned mean power ratio was 1.710, with a 95% interval of 0.160 to 4.750. One large value controls much of that mean. The ratio is unchanged by a constant gain on each complete pair signal. Labels were used only to score the selections. This is a reused epilepsy cohort, with no BCI task or stimulation test. The package preserves code, source versions, retrieval checks, compact data, patient outcomes, and figures.

stereoelectroencephalography · SEEG · epilepsy · seizure onset · gamma power · beta power · power ratio · channel selection · brain-computer interface · BCI · signal processing · patient bootstrap