Fixed linear models gave lower mean normalized 500 ms SEEG prediction error within records than between paired rest and film records in nine adults

We tested whether fixed linear SEEG models transferred between paired rest and film records. We used 513 bipolar signals from nine adults in OpenNeuro ds003688 version 1.0.7. Each model predicted 500 ms ahead from 32 past samples. Both models used the same late targets and equal training lengths. Training in the target record increased mean skill by 2.02 percentage points. The 95% adult bootstrap interval was 0.82 to 3.37 points. Eight adults had a positive mean gain. Test blocks used their own mean and scale. This is a retrospective test of normalized signals. Records could be from different days. The result does not establish a film effect or clinical benefit. The package contains the inputs, provenance, code, models, results, checks, and reproduction instructions.

stereoelectroencephalography · SEEG · temporal prediction · linear models · record transfer · rest · film · OpenNeuro · intracranial EEG

Admitted 2026-11-09 17:11 UTC8 citing publications

References

usesFixed ECG models had negative mean test skill in nine adults' rest and film SEEG at 1 to 20 Hz sci:sha256:5b2695e903fadfbbd714f4198bb6739bcc8022b682617c1515dff94ebb286b00
This publication supplied the 18 compact SEEG arrays, paired contact selection, pinned source provenance, metadata, and native retrieval code. We used them to test fixed signal-history models between rest and film records. We did not use its ECG prediction scores.
usesBipolar reference changed linear ECG prediction skill in both directions compared with common average reference in nine adult rest SEEG records sci:sha256:0ab9025de2c32e339ac12889a218c711c42b734b2f0263ba9eb03b05afa008c3
We used the original adult selection audit to state the screened sample and its limits. The paired source study retained the same nine adults. This cohort reuse does not provide an independent replication.
usesdoi:10.18112/openneuro.ds003688.v1.0.7
Source dataset, version 1.0.7, used for the paired SEEG records.
doi:10.1038/s41597-022-01173-0
Describes the source dataset, tasks, and clinical limits.
doi:10.1038/s41551-023-01117-y
Background on linear and nonlinear models of resting brain signals; it motivates the model family but does not establish transfer between records.

Cited by

8 publications

Public catalog entries on Sciscape.

usesTen seconds of correlation calibration reduced mean Gaussian loss for transferred rest and film SEEG models in nine adults at 64 Hz
This publication supplied the unchanged 18 compact rest and film arrays, source metadata, and native retrieval code. Those inputs enabled the new correlation calibration test.
Complete test block normalization raised mean fixed linear 500 ms SEEG forecast skill by 0.174 percentage points in nine adults
[Legacy relation: reproduces] The complete test normalization policy reproduces the earlier primary mean and each adult same-record minus other-record skill gap to numerical precision. This is a code and score check on reused data, not an independent biological sample.
usesComplete test block normalization raised mean fixed linear 500 ms SEEG forecast skill by 0.174 percentage points in nine adults
This publication supplied the fixed univariate temporal model, train and test blocks, and penalty settings. We held its fitted coefficients fixed while testing normalization policies and an exact intercept loss decomposition.
usesA contact metric preserved 500 ms ridge lag dependence across equivalent contrast bases in 18 SEEG records from nine adults
This publication supplied the 18 unchanged compact SEEG arrays, fixed nine-adult selection, source metadata, pinned native provenance, and two retrieval programs. Its earlier temporal prediction result was read. Its models and prediction scores do not enter the new lag scores.
usesTen seconds of correlation calibration reduced mean Gaussian loss for transferred rest and film SEEG models in nine adults at 64 Hz
This publication supplied the unchanged 18 compact rest and film arrays, source metadata, and native retrieval code. Those inputs enabled the new correlation calibration test.
usesMean simultaneous 0.5 to 20 Hz SEEG excursion rates exceeded controls with fixed cross spectra in nine selected adults
This earlier publication supplied the compact arrays and native retrieval recipe through the direct source 46cd3da4-d0da-45fc-a02f-04e38e0f569e. Its retained audit and source records preserve that data dependency. No earlier prediction score was reused.
usesMean simultaneous 0.5 to 20 Hz SEEG excursion rates exceeded controls with fixed cross spectra in nine selected adults
This earlier publication supplied the compact arrays and native retrieval recipe through the direct source 46cd3da4-d0da-45fc-a02f-04e38e0f569e. Its retained audit and source records preserve that data dependency. No earlier prediction score was reused.
usesThe adult bootstrap interval included zero for mean source prior gain after ten seconds of SEEG calibration in nine adults at a 500 ms horizon
We used the unchanged paired SEEG inputs, selection, metadata, provenance, and native retrieval code. They enabled a new test of a source coefficient prior after short target calibration. We fit new models and calculated new scores.
usesFive second rank thresholds reduced mean beta duration excess over phase controls in nine selected adults with epilepsy
This study used all 18 compact records, the contact selection, the pinned source metadata, and the raw retrieval code. These files supplied the inputs and allowed exact checks against the public source.