Five SEEG sessions did not establish higher site accuracy for four-pulse voltage averaging than uniform-tie voting at 20–100 ms

We tested whether averaging four SEEG responses identifies a stimulation site better than voting on separate pulse labels. OpenNeuro ds004457 version 1.0.2 supplied five clinical sessions and 185 eligible sites. Templates used earlier pulses at each site. Tests used identical final four-pulse groups and fixed 20–100 ms features. Mean voltage accuracy was 28.49%. Expected uniform-tie voting accuracy was 26.85%. The paired mean difference was +1.64 percentage points. Its descriptive 95% patient bootstrap interval was −1.27 to +5.48 points. The exact paired test gave p = 0.5625. These five sessions did not establish an averaging advantage. We also proved an exact limit: uniform two-pulse voting has the same expected accuracy as the separate predictions. Recorded four-pulse groups had a median span of 14.19 seconds. The package preserves three figures, code, derived data, source hashes, a retrieval recipe, and full calculation checks. This offline test requires known pulse times and an unchanged site. It does not prove neural origin or online BCI performance.

stereoelectroencephalography · SEEG · electrical stimulation · brain-computer interface · site identification · repeated pulses · voltage averaging · voting · tie handling