A five-pulse stability gate selected choices with less observed later score loss in eight SEEG participants at 100 to 500 ms under a half-energy rule

We tested whether a stability gate can identify reliable stimulation-site choices from five calibration pulses. We used public STReEF SEEG data from eight people with focal epilepsy: 31 candidate groups and 258 target tasks. The site rule selects the largest target-to-other-shaft score among candidates with at least half the maximum target energy. The primary window is 100 to 500 ms. The gate accepts a choice only if removing any one calibration pulse leaves the choice unchanged. It accepted 149 targets. On five separate later pulses with reversed recorded contact order, accepted choices had 0.692 percentage points less observed spatial score loss than uniform random target acceptance at the same count. Each group and person has equal weight. The descriptive 95% person bootstrap interval was 0.051 to 1.278 points; the two-sided sign-flip value was 0.09375. This does not establish a reliable population benefit. A secondary later target-energy failure rate was 7.11% for accepted choices and 15.05% for the control. The earlier 20 to 100 ms window did not show the same benefit. The gate rejected 42% of targets. Time and recorded contact order change together, and most candidate test pulses precede completion of group calibration. This is an offline diagnostic association, not a clinical or prospective device test. The package preserves exact pulse products, all choices, figures, source code, data pins, and repeat instructions. Fresh source checks verified all 960 raw ranges. Independent arithmetic, split checks, and a repeated analysis passed.

stereoelectroencephalography · brain-computer interface · single-pulse stimulation · stability gate · site selection · target acceptance · response score · focal epilepsy