A fixed high gamma to beta power ratio increased drift in marked windows after stimulation in 103 visual SEEG records from 19 people

We tested a fixed high gamma to beta power ratio in public visual SEEG data. The question was whether the ratio reduced feature drift after cortical stimulation. The primary test used 103 records, 10,211 channel-record pairs, and 19 people from OpenNeuro ds005169 version 1.0.0. The ratio increased median absolute drift in all 19 people. The median paired increase was 0.767 dB. The 95% person bootstrap interval was 0.655 to 0.848 dB. The exact two-sided sign test gave p = 0.00000381. The baseline increase was 0.755 dB. The interval for the post-minus-baseline comparison included zero. The ratio cancels a shared voltage gain that is constant within a window. This property did not make the tested feature more stable. The result does not show that stimulation caused the increase. Decoder accuracy was not tested. The package preserves all 112 record spectra, source hashes, original metadata, code, tables, bootstrap draws, four figures, and reproduction instructions. All 336 source file checks passed. A fresh rebuild gave identical arrays and results. This is an exploratory follow-up in a reused cohort.

stereoelectroencephalography · SEEG · high gamma · beta power · feature drift · brain-computer interface · Welch spectrum