Conditional Gaussian shaft scores had lower mean file rank agreement than marginal scores in 36 SEEG patients at 1–80 Hz with 10% covariance shrinkage

We tested whether control for other electrode shafts increases agreement of SEEG connection ranks across two interictal files. The study used 36 patients from OpenNeuro ds004100 version 1.1.3. Both Gaussian scores used the same local contact contrast spaces. At 1–80 Hz with 10% covariance shrinkage, mean file rank agreement was 0.8765 for marginal scores and 0.8516 for conditional scores. The mean paired difference was −0.0249. Its 95% patient bootstrap interval was −0.0492 to −0.0022. The interval assumes independent, exchangeable patients. All fixed checks had negative mean differences, but two intervals included zero. The strongest edge list did not show an established retention gain. All 72 input covariance archives matched new builds from pinned public bytes exactly. A second EDF reader checked all 72 source ranges. The package preserves code, covariance data, source versions, range hashes, every score, all uncertainty draws, and repeat instructions. This result concerns the tested short windows and score rule. Stability does not establish neural or clinical validity.

stereoelectroencephalography · intracranial EEG · functional connectivity · conditional dependence · file rank agreement · covariance shrinkage