FRandR onset marks had more 10 ms coincidences than a 1 s shift model predicted across pairs with no shared contact in nine sleep SEEG patients

We tested FRandR onset coincidence in 153 five-minute sleep records from nine temporal lobe epilepsy patients in OpenNeuro ds003498 version 1.0.1. The primary channel pairs shared no contact. Independent circular shifts held each channel count and pattern fixed within one-second blocks. The mean patient coincidence probability was 10.36%, against an expectation of 2.05%. The mean excess was 8.31 percentage points. The 95% patient bootstrap interval was 5.39 to 11.41 percentage points. All nine patient effects were positive. None of 4,999 joint shift draws reached the observed group statistic; the one-sided Monte Carlo p value was 0.0002. Fixed checks with smaller blocks and different shaft names also had positive group effects. The test rejects the specified rotation model. It does not establish direct neural coupling or clinical prediction. The source paper identifies depth electrodes, but converted metadata labels them ECOG. This is an event annotation study. A separate method matched all 1,590 record and condition counts. A fresh public download produced identical core result files. The package preserves source tables, hashes, code, results, checks, and run instructions.

FRandR · high-frequency oscillations · intracranial EEG · sleep · event timing · coincidence · conditional resampling · temporal lobe epilepsy