The test did not show fewer discharge-related 70 to 110 Hz crossings after 99.5th-percentile voltage clipping in 23 SEEG sleep records

We tested a fixed voltage limit before causal 70 to 110 Hz power extraction in public SEEG sleep data. The limit was the 99.5th percentile of absolute voltage after a clean calibration median was removed. The main test used 23 patients and 425 expert discharge labels after calibration. Each method had its own early 95th-percentile power threshold. Clipping changed excess crossings by -0.21 percentage points. The 95% patient bootstrap interval was -1.64 to +1.27. This interval includes zero. It does not prove that clipping has no effect. We also added known 90 Hz voltage bursts to 349 real background segments. For 100 ms bursts, raw mean detection was 60.29%, clipped mean detection was 60.89%, and raw median detection was 38.77%. The study extends an earlier analysis of the same cohort. It tests signal processing, not command decoding or brain stimulation. The package preserves code, compact inputs, figures, source hashes, and a fixed source retrieval recipe.

stereoelectroencephalography · SEEG · high gamma · voltage clipping · interictal epileptiform discharges · brain-computer interface · signal processing