Models with zero covariance between recorded contacts gave mean overlap of 81.7 percent for top 10 percent bipolar SEEG hubs in paired 60-second samples from 36 patients at 1 to 80 Hz

We tested whether top-strength bipolar SEEG hubs can repeat without covariance between different recorded contacts. We used paired 60-second samples from 36 epilepsy patients in public OpenNeuro dataset ds004100, version 1.1.3. The model kept each contact variance and set all other contact covariance to zero. At 1 to 80 Hz, the top 10 percent model hubs had mean overlap of 81.74 percent across records. The descriptive 95 percent patient bootstrap interval was 78.04 to 85.35 percent. A control shuffled variance positions within each shaft. It gave 18.86 percent overlap. The paired mean difference was 62.88 percentage points, with a 95 percent interval of 58.73 to 66.65 points. The difference was positive in all 36 patients. Model hubs matched measured hubs by only 4.75 and 7.52 percent in the two records. Thus, high repeatability alone does not establish covariance between recorded contacts. The result does not identify false measured hubs or estimate an artifact fraction. The package preserves inputs, source retrieval records, code, scores, and all control scores. Fresh public data checks passed for all 72 records. Independent calculation checks passed. Two complete numerical repeats matched all 41 output files.

stereoelectroencephalography · bipolar montage · functional connectivity · hub overlap · contact covariance · measurement model