A 24-pulse screen had a 2.0-point higher mean response fraction in eight-site SEEG tests from five patients when stored responses were assumed to stay fixed

We tested an early screen for stimulation site selection with public SEEG data from five patients. Each test had eight candidate sites. The rules assumed that stored responses would stay fixed if pulses were omitted or put in a different order. The staged rule used two pulses at each site, then two more at each of four retained sites. Both it and the three-pulse-per-site reference used 24 fit inputs. Their mean later response fractions were 86.72% and 84.68%. The gain was 2.04 percentage points. The descriptive 95% patient bootstrap interval was -0.10 to 4.16 points. The sign flip p value was 0.1875. These data do not establish a gain in the patient population. In one patient, the screen removed the site with the largest measured later response in 76.2% of pools. No pool had all fit pulses before all test pulses across sites. The package has source epochs, code, two figures, all pool results, hashes, and raw retrieval steps. A fresh raw rebuild matched all nine epoch files exactly. The endpoint measures voltage response, with no test of clinical benefit or live interface performance.

stereoelectroencephalography · brain-computer interfaces · electrical stimulation · site selection · pulse allocation · evoked voltage