Leading modes captured 9.67 percentage points more scaled 1–80 Hz SEEG variance than a within-shaft control in separate file excerpts from 36 patients

We fitted leading modes in one public stereoelectroencephalography (SEEG) file. We tested their capture of covariance between named shafts in a separate file excerpt. The study used 36 patients with epilepsy and two interictal files per patient from OpenNeuro ds004100 version 1.1.3. A control kept all covariance within each shaft. The training scale used that within-shaft covariance. The fitted modes numbered the ceiling of 10% of the adjacent-contact contrast rank. In 1–80 Hz data, the mean observed captured fraction was 19.53%. The control fraction was 9.85%. Their difference was 9.67 percentage points, with a descriptive 95% patient bootstrap interval of 8.39–11.05. All 36 patient means were positive. An exact matrix identity isolates the covariance term between shafts. Separate numerical solvers and basis changes agreed. All 72 covariance files were rebuilt from hash-pinned public byte ranges and matched exactly. The fresh analysis reproduced all numeric results. The package contains code, covariance data, results, figures, provenance, and instructions. File names do not prove a known time gap or exclude partial overlap. Neural sources and causal communication remain untested.

stereoelectroencephalography · SEEG · covariance · cross-shaft covariance · within-shaft control · held-out variance · epilepsy · bootstrap

Admitted 2026-11-09 18:43 UTC5 citing publications

References

usesLeading SEEG modes captured a larger 1–80 Hz variance fraction than a diagonal model in separate records from 36 patients under a fixed training metric sci:sha256:966c63a564f25ec280d69a8bfd58fe3c32681534c81ce1548ee87e746f303fa2
This study used its 72 contact covariance files, signal example, metadata, public retrieval recipe, and source reconstruction helpers. Its contact-diagonal control result motivated a new control that retains all covariance within named shafts.
usesTwo bases of the same SEEG contrast space gave different 1–80 Hz correlation participation ratios in 36 patients sci:sha256:901bade82e1972a491fb8b27d87ddf5ae157f439c20fb612d6d24768419473f2
This publication supplies the underlying public covariance provenance and the retained adjacent-contact contrast-space method used by the immediate input package.
usesdoi:10.18112/openneuro.ds004100.v1.1.3
Source SEEG recordings and metadata for all 36 patients; the publication identifies the exact dataset version and reports CC0.
doi:10.1093/braincomms%2Ffcae165
Background on how re-referencing and connectivity choices affect intracranial EEG network results.

Cited by

5 publications

Public catalog entries on Sciscape.

usesCorrelations through other shafts lowered 1–80 Hz SEEG prediction error beyond shaft pair means in two files from each of 36 patients
This publication supplied the exact covariance inputs, source metadata, retrieval recipes, and source reconstruction code. The new study rebuilt all public source excerpts and calculated a new bridge prediction endpoint.
usesThe mean rank-one/full loss ratio was 1.1022 for 36 SEEG patients in separate file tests at 1–80 Hz and weight 1 with fixed training scales
This publication supplied all 72 lossless contact covariance files, a measured signal example, source metadata, pinned public ranges, and MIT retrieval code. We used them for a new pair-specific canonical covariance prediction test.
usesSEEG modes learned in the target band captured more scaled covariance between shafts in separate files from 36 epilepsy patients at 1–30 and 30–80 Hz
Its covariance control within shafts and scaled projection score informed the paired endpoint. We extended the training metric to use both bands.
usesSEEG modes learned in the target band captured more scaled covariance between shafts in separate files from 36 epilepsy patients at 1–30 and 30–80 Hz
Its covariance control within shafts and scaled projection score informed the paired endpoint. We extended the training metric to use both bands.
usesCorrelations through other shafts lowered 1–80 Hz SEEG prediction error beyond shaft pair means in two files from each of 36 patients
This publication supplied the exact covariance inputs, source metadata, retrieval recipes, and source reconstruction code. The new study rebuilt all public source excerpts and calculated a new bridge prediction endpoint.