A unit covariance load reduced beta excess coherence in the next minute in 12 of 19 selected SEEG records

We tested if a fixed unit covariance load improved multichannel beta coherence in the next minute. The test used 19 selected public SEEG records from OpenNeuro ds003876 version 1.0.2. Each fit used 30 two-second segments. The next 30 segments tested the fixed complex weights. We subtracted the exact coherence expectation under independent phase rotations with amplitudes fixed. The unit load reduced this excess score in 12 patients and increased it in seven. The equal-patient mean difference was 0.003360. The median was -0.002140. The descriptive 95 percent patient bootstrap interval was -0.004497 to 0.016326. The planned two-sided sign reversal p value was 0.951271, under independent patients and symmetric null signs. Omission of one large positive patient changed the mean to -0.002487. The fixed load gave no consistent gain in this sample. Fresh source downloads rebuilt all 19 inputs exactly. Separate solvers and two complete replays checked the results. Code, inputs, weights, source hashes, and instructions are preserved. The result applies to these records and this complex estimator. It does not establish causal connectivity or clinical value.

stereoelectroencephalography · SEEG · coherence · covariance regularization · phase reference · epilepsy