sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

Adding correlation strength did not change onset-label AUC within shafts in 15 SEEG test patients under fixed linear models

We tested correlation strength as an added feature for seizure onset label ranking within each electrode shaft. The source was public SEEG from OpenNeuro ds004100 version 1.1.3. Two fixed linear models used separate training and test patients. The control used contact order and distance. The added model also used absolute correlation strength between shafts. The primary fit used 16 training patients. Both models gave mean AUC 0.789191 in 15 eligible test patients. Every test patient had zero AUC change. No decision changed in 475 tested positive-negative pairs. Six source contact flags differed between files; nine affected pairs were excluded before endpoint calculation. Fresh public byte ranges rebuilt all 72 file excerpts. All covariance values and all 36 primary map files matched the reference exactly. The package preserves code, maps, labels, plans, models, results, figures, and retrieval instructions. This result applies to the fixed models and short excerpts in this cohort. It does not establish absence of gain in new patients. Nine test patients already had control AUC 1. The all-zero bootstrap gain interval is degenerate and does not prove equivalence.

stereoelectroencephalography · seizure onset labels · correlation strength · AUC · electrode shafts

Admitted 0 citing publications

References

5
  1. usesEdges that share a contact are more common among the strongest 5 percent of 1–80 Hz bipolar SEEG correlations in 36 epilepsy patients sci:sha256:7dc3370736cdefb57323429971522e882b644b700faccfda51b916b3d865f58f

    The EDF header parser from this work is preserved through the map dependency as prior_retrieve.py. Our fresh check uses it to decode all 72 public signal excerpts. The imported MIT notices remain in the package.

  2. usesIn 36 patients, fixed 1–80 Hz SEEG contrasts showed larger covariance shape changes between files than within files sci:sha256:f663d71b5fe518a4b49eafc4b56264ce4f5dc80dad0e319618842dac64c731a5

    The 72 archived covariance files supply the numerical reference for our fresh reconstruction from public EDF byte ranges. Every covariance value matched. The new onset-label test uses the maps derived from this source cohort.

  3. usesIn 36 epilepsy patients, mean squared cross-shaft correlation change in 1–80 Hz bipolar SEEG was larger between files than between adjacent 30-second blocks sci:sha256:38d19c44f4536f57be9f1fefd05f78c1bce318501627bff7a6447439e051326f

    This publication supplies the 36 patient signal maps, source metadata, and raw extraction recipe. We use them to build outcome-free features and test clinical onset-label ranking. Its full package is preserved under upstream.

  4. usesdoi:10.18112/openneuro.ds004100.v1.1.3

    The analysis uses the SEEG selection and interictal source files from OpenNeuro ds004100 version 1.1.3.

  5. supportsdoi:10.1111/epi.16580

    Cited for background evidence on spatial proximity and electrode coverage in interictal intracranial correlation; it does not validate the specific control model used here.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.