Same site templates had lower centered late posterior cingulate SEEG error than other sites in six epilepsy patients with fixed gain and baseline control

We tested fixed voltage prediction after hippocampal complex stimulation in public OpenNeuro ds004696 version 1.0.1. Six selected patients had 14 posterior cingulate contacts, 19 source pairs, and 190 distinct pulses. The first five pulses formed a template. The next five formed a test mean. We centered each window and fixed the prediction gain at one. At 100 to 500 ms, same site templates had lower error than other site templates in all six patients. After a before pulse control, the mean gain was 5.285 times the training baseline energy. The 95% patient bootstrap interval was 2.315 to 8.496. The exact two sided sign p value was 0.03125, under independent patients and symmetric null signs. The average patient after error reduction against other sites was 68.78%. A later descriptive control also favored same site templates over zero voltage in all six patients. Fresh raw reconstruction matched all 1,881,408 voltage values. Independent code confirmed all seven analyses. Same site pulses occur in one clinical block, so brain state and artifacts can produce site differences. This follow-up uses the parent sample and does not establish a direct pathway or clinical biomarker. The package preserves data, code, source versions, results, and reproduction instructions.

stereoelectroencephalography · hippocampal stimulation · posterior cingulate cortex · voltage prediction · source specificity · epilepsy