With a 0.8 correlation cutoff, means of five later pulses matched early templates more often than one later pulse in 196 selected SEEG readouts from eight patient sessions

We tested whether a repeatable SEEG response averaged across five pulses also gave a repeatable response after one pulse. We used public limbic stimulation data from all eight patients with epilepsy in OpenNeuro ds004696 version 1.0.1. The first five good pulses selected 196 bipolar readouts at 66 stimulation sites. The final five good pulses were a separate test set. A waveform matched the early template when Pearson correlation was at least 0.8 over 20 to 500 ms. Each patient had equal weight. Individual later pulses matched in 66.4% of cases. Means of five later pulses matched in 94.5% of cases. The gain was 28.2 percentage points. The patient bootstrap 95% interval was 16.6 to 39.9 points. The exact two-sided sign-flip test gave p = 0.00781. The gain was positive in all eight patients. All five individual later pulses matched at 41.1% of selected readouts, with equal patient weight. The gap remained with equal site weights and different correlation cutoffs. This result extends an earlier test of five-pulse averages in the same records. It does not establish neural origin, online BCI control, or stability across days. The package supplies exact trial products, predictions, source code, three figures, source hashes, and reproduction instructions. A new complete source hash check and extraction for one patient reproduced the products within numerical precision. The other seven patients use preserved products from the earlier publication.

SEEG · intracranial stimulation · single-pulse responses · signal averaging · response repeatability · brain-computer interfaces