25 exact STE HFO checks rejected 9.17 percent of uniform pair label paths on average across 18 SEEG records

We tested repeated exact HFO count checks with public SEEG marks from OpenNeuro ds005398 version 1.1.3. The primary set had 308 STE channels from 18 sleep records. Each channel had at least 20 first-hour marks. We held the counts fixed and assigned independent uniform labels within 30 adjacent minute pairs. Integer path counts gave the exact chance of any rejection across 25 checks. The equal patient mean was 9.17 percent for repeated tests at 0.05. One final test gave 3.23 percent. Bonferroni gave 0.31 percent. An exact error-budget rule gave 3.72 percent. Every corrected channel stayed below five percent. Shared label draws also checked each patient's channel set. The package preserves source bytes, versions, code, all boundaries, results, a proof, and reproduction instructions. Independent checks passed. An offline repeat gave identical bytes for 12 scientific output files. These assigned label results do not establish a neural effect. A natural contrast requires a valid sign reference.

high-frequency oscillations · SEEG · sequential testing · exact calibration · error spending