The patient median finite-trial energy fraction was 80.7% for random eight-pulse SEEG averages at 50 to 250 ms in six patients after a 500 microvolt gate

We measured how trial variation affects mean evoked SEEG energy from 50 to 250 ms. We used 16 probe files from six patients in OpenNeuro ds006065, version 1.0.0. A 500 microvolt gate and a 16-trial minimum kept 15 files and seven pre/post pairs. Pulse artifacts supported two-second epoch reconstruction because public event markers were absent. For random eight-pulse subsets, the patient median finite-trial energy fraction was 80.7%. The 95% patient bootstrap range was 25.5% to 97.2%. Six full-file corrected estimates were negative. Correction kept the change sign in all seven primary pairs. Fixed bipolar, stricter-gate, and ungated checks gave medians of 82.3%, 54.7%, and 87.3%. The subset identity is exact for the observed trial pools. Its interpretation as mean response energy requires a common trial mean and zero between-trial residual covariance. This selected patient cohort does not establish a clinical effect. The package contains signals, source versions and hashes, code, figures, quality masks, and repeat instructions. A fresh-process repeat matched all five numerical output files.

SEEG · evoked response · finite-trial variance · signal energy · hippocampus · trial averaging · intracranial EEG · measurement study