Ordered 20–100 ms SEEG bins gave higher site accuracy than sorted bins in five sessions with a fixed reader

We tested whether time positions help read stimulation sites beyond each recording pair's centered signed voltage distribution. We used public OpenNeuro ds004457 version 1.0.2. Five SEEG sessions had 183 sites and 732 test pulses. The first four eligible pulses per site fit each reader. The final four tested it. At 20–100 ms, ordered bins gave 40.56% mean site accuracy. Ascending sorted bins gave 27.22%, with the same 16 signed values per pair. Each patient had equal weight. The mean difference was 13.33 percentage points and was positive in every patient. The descriptive 95% patient resample interval was 10.71 to 16.00 points. The exact two-sided sign-flip reference gave p = 0.0625. Two other fixed readers also gave positive gaps in every patient. The before-pulse mean gap was −0.13 points. This is an exploratory comparison in a reused cohort. Site blocks, retrospective pulse selection, and possible artefact limit interpretation. The result applies to these readers and does not establish neural origin, online control, or a population effect. The package preserves compact data, all planned scores, predictions, fitted states, code, provenance, and three figures. Fresh raw reconstruction and exact score replay passed.

SEEG · stimulation · temporal order · site classification · sorted voltage bins