Distance had an uncertain adjusted link to later variance relative to baseline in 20–500 ms SEEG from eight people

We tested distance and later pulse variance in public SEEG data from eight people with epilepsy. The data contained 886 stimulation-to-readout pairs from 86 stimulation sites. We used the first five good pulses to measure response energy. We used the final five distinct pulses to measure variance. The post-pulse window was 20–500 ms. The outcome was the log ratio of post-pulse variance to baseline variance. Within each person, we controlled stimulation site, readout pair, and ranked early response energy. The equal-person mean adjusted rank correlation was -0.069. The 95% person bootstrap interval was [-0.216, 0.085]. The interval did not resolve the direction of the adjusted mean association. The exact person sign-flip reference was p = 0.4297. A planned model with separate baseline control gave mean correlation -0.208, with interval [-0.283, -0.124]. The conclusion thus depends on the baseline model. This is an exploratory association test. It does not establish a cause or a gain in BCI performance. The package contains source hashes, trial products, geometry, figures, code, and a fresh source waveform check.

stereoelectroencephalography · SEEG · brain-computer interface · electrical stimulation · response variance · electrode distance · rank correlation · human limbic system