A rolling threshold reduced the mean SEEG call gap in eight patients with fixed templates and 20 to 500 ms windows

We tested whether five recent past controls improve response calls over five frozen early controls. The test used fixed signed templates, 20 to 500 ms SEEG windows, 610 later pulses, 86 selected readouts, and eight reused patient sessions from OpenNeuro ds004696 version 1.0.1. The equal patient mean after-minus-before call gap fell from 74.34 to 70.98 percentage points. The rolling-minus-frozen change was -3.36 points. Its descriptive 95% patient bootstrap interval was -5.53 to -0.83 points. Six patients had decreases. The sign-flip reference gave p = 0.046875. After calls stayed near 88.2%, while before calls rose from 13.88% to 17.23%. The 100 to 500 ms check did not show a clear decrease. Two nearly simultaneous event rows were excluded from the primary replay after a metadata audit. A local plan, an amendment, and a timing note are preserved. This is an exploratory offline method result. It does not establish neural specificity or clinical benefit. The package preserves exact input voltage, source indices, metadata, all scores and threshold contributors, templates, four figures, code, and retrieval recipes. Independent arithmetic checked every score and call. A clean run reproduced the numerical outputs. Fresh checks covered two complete public source files and all 44 metadata files. Checks for the other six source files are identified as inherited.

stereoelectroencephalography · SEEG · brain-computer interface · electrical stimulation · response detection · rolling threshold · signal processing