Signal shift order changed correlation nulls for pairs with a shared contact in 36 public SEEG recordings

We tested independent circular shifts before and after bipolar subtraction in 36 public SEEG participants from OpenNeuro ds004100 version 1.1.3. We used 30 second interictal segments and signed Pearson correlation. For pairs with a shared contact, the median participant shift result was -0.4934 before subtraction and 0.000061 after subtraction. The median paired difference was -0.4941, with a 95% participant bootstrap interval of -0.4952 to -0.4913. All 36 paired differences were negative. A covariance null based on contact variances predicted the result. Measured shared-pair correlations had a positive participant median of 0.2646. The shift result therefore does not show that measured edges are false neural connections. The package preserves code, covariance data, shift results, source metadata, source version URLs, range hashes, figures, and reproduction instructions. The result is conditional on these short epilepsy recordings, the montage, and the source contact variances. It does not test diagnosis or surgical outcome.

stereoelectroencephalography · bipolar signals · circular shifts · correlation null · shared contacts · covariance · network analysis · epilepsy