Spatial variation of overlapping HFO marks was 1.15 times the mean from cluster exchange within size groups in 134 SEEG sleep records from eight patients

This study tested spatial count variation in public SEEG sleep records. The sample had 134 eligible five-minute records from eight patients in OpenNeuro ds003498 version 1.1.1. The primary marks were ripples that overlapped fast ripples. A new model kept total counts in each 30-second part and the set of short cluster pair-count patterns fixed. It exchanged patterns only between clusters with the same size. The normalized statistic had an exact model mean of 1. The observed equal-patient mean was 1.147. None of 9999 conditional draws reached it. The added-one Monte Carlo p value was 0.0001. Only three of eight patient means exceeded 1. The patient bootstrap interval was 0.963 to 1.445. Removing one patient with one record reduced the mean to 1.007. The 10-second analysis gave a mean of 0.998 and a conditional p value of 0.528. The primary sample rejected the specified exchange model. Evidence for a common effect across patients was limited. The package contains the model proof, source files, code, all planned checks, saved draws, and figures. An independent recount passed. Full offline results matched byte for byte. Fresh public source retrieval passed. This is a new secondary analysis of the source cohort, not an independent replication.

high frequency oscillations · intracranial EEG · spatial count variation · conditional randomization · epilepsy · stereoelectroencephalography · cluster exchange