Ripple rate lists from 60 seconds had 13.7 percentage points lower mean agreement with overlap-HFO lists from other records in seven SEEG patients, with equal patient weights

We tested whether frequent ripple marks could replace overlap-HFO marks for short-sample rate lists. We used OpenNeuro ds003498 version 1.1.1. The fixed main analysis used 135 sleep SEEG records from seven patients. Each source list used 60 seconds. Its target used all 300 seconds of a different record from the same patient. Lists used the highest 20 percent of rates, with list size rounded up. Tied counts had equal fractional membership. Each patient had equal weight. Ripple agreement was 51.17 percent. Direct overlap-HFO agreement was 64.90 percent. The mean difference was minus 13.72 percentage points. The exploratory 95 percent patient bootstrap interval was minus 32.79 to plus 1.45 points. The exact two-sided sign-flip p value was 0.265625. Five patient differences were negative and two were positive. The primary result does not establish a population difference or equal performance. Ripple marks outnumbered overlap-HFO marks by 14.81 to 1. Longer source samples retained a negative mean difference. The result shows that mark abundance alone did not give a closer average match to this target. It measures agreement with released detector lists. It does not measure clinical accuracy. The package preserves source event tables, pinned versions and hashes, code, the fixed local plan, counts, patient results, figures, and reproduction instructions. Independent counts, scalar scores, exact resampling checks, and a full offline repeat passed. A fresh public source check matched all 185 event tables. This is a new secondary analysis of an existing cohort.

stereoelectroencephalography · high frequency oscillations · ripples · event lists · repeatability · secondary analysis