SEEG error products from 20 to 100 ms exceeded a before-pulse control in four of five offline 6 mA sessions

We tested whether an error in a stimulation response map repeats across pulse groups. We used public SEEG data from five epilepsy patients in ds004457, version 1.0.2. At 52 middle sites, we estimated the response from two flanking sites. We multiplied residuals from the first four and final four good pulses. Under stable response means and zero cross-group error covariance, this product estimates squared repeatable error. The 20 to 100 ms score exceeded a before-pulse control in four of five sessions. The mean contrast was 1,374.71 microvolts squared. The median was 8.87. The patient sign-flip p value was 0.3125. A fixed check with readout pairs that share no contacts reduced the mean to 37.90 microvolts squared. One readout pair and early time bins had large effects. The patient test did not establish a common effect across sessions. The package provides an audit method, three figures, all scores, compact data, and raw retrieval code. A raw rebuild matched 2,451,400 finite bins. The study did not test brain-computer interface performance.

SEEG · stimulation response maps · cross-validated error · repeatability · intracranial EEG · pulse groups