Early channel corrections reduced mean later detector timing error across 18 SEEG sleep records at a 20 ms match limit

We tested early channel corrections on later MNI and STE detector marks in public SEEG sleep data. The source was OpenNeuro ds005398, version 1.1.3. We fixed within-channel matches with start differences at most 20 ms. We fitted corrections in the first half hour and scored them in the next half hour. For 2,877 later pairs on 180 channels in 18 patients, channel corrections reduced mean absolute timing error. The equal-patient errors were 4.808699 ms for channel corrections and 5.750928 ms for one patient correction. The mean gain was 0.942228 ms, with a descriptive 95 percent patient bootstrap interval of 0.422405 to 1.504431 ms. Fourteen patients improved, two had larger errors, and two had no change. The gain was absent in the subset that required at least 20 early matches per channel. The secondary onset order check did not establish a consistent gain. Population inference requires independent, exchangeable patients; the source does not establish these assumptions. The local plan followed earlier cohort findings and was not externally registered. This study measures agreement between released detector outputs, not true neural times or clinical benefit. The package preserves exact public inputs, retrieval pins, code, event scores, figures, and reproduction instructions. Independent checks and two full rebuilds passed.

stereoelectroencephalography · high-frequency oscillations · detector timing · channel calibration · event matching · intracranial EEG