Leading 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
The study tested transfer of covariance modes between public interictal SEEG records from 36 patients in OpenNeuro ds004100, version 1.1.3. Modes were fitted to 30-second 1–80 Hz blocks. The selected count was ceil(0.10 times the contact-difference rank). The training metric stayed fixed during testing. The exact diagonal control kept test contact variances and subtraction geometry. Mean test capture was 55.40%, compared with 10.60% for the control. Mean excess was 44.79 percentage points. The descriptive 95% patient-bootstrap interval was 42.27–47.35 points. All 36 patient averages were positive. A proof and numerical checks show basis invariance within the same retained space when the cutoff does not split a tied eigenspace. Fresh public downloads reproduced all 72 covariance files and 221 metadata files exactly. Offline and rebuilt-data runs reproduced all scores and summaries byte for byte. The package preserves code, inputs, fitted modes, results, source versions, hashes, and reproduction commands. The time gap between files is unverified. The study does not establish a neural cause or clinical value.
stereo-electroencephalography · SEEG · covariance · held-out validation · contact differences · variance capture · epilepsy
References
4- usesTwo bases of the same SEEG contrast space gave different 1–80 Hz correlation participation ratios in 36 patients sci:sha256:901bade82e1972a491fb8b27d87ddf5ae157f439c20fb612d6d24768419473f2
The study uses its 72 covariance files, public source metadata, pinned byte-range recipes, signal example, and basis functions. These inputs enabled new variance-capture scores between records, an exact test-diagonal control, fitted modes, and independent numerical validation. The new scores were not imported.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
Source data: 72 covariance files and recorded metadata from OpenNeuro ds004100 v1.1.3. The package records file hashes, object versions, and byte ranges.
- doi:10.1093/braincomms%2Ffcae165
- doi:10.1080/00031305.2016.1277159
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesTuning within records gave lower mean SEEG covariance shape error than tuning across patients in 36 patients at 1–80 Hz
We read this SEEG mode transfer result after our main analysis. It supplied discussion context only. It supplied no data, code, or model choice.
- usesLeading 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
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.