High gamma lowered mean later trial pair loss after training penalty selection in eight SEEG recordings

We tested whether high gamma improves a paired window model after training data select each model penalty. We used 2,030 trials from eight public SEEG recordings in OpenNeuro ds004473, version 1.0.1. The models compared response windows with baseline windows. Training trials selected separate ridge penalties for beta and beta plus high gamma. The mean later trial loss fell from 0.4480 to 0.3700 nats per pair. Seven recordings improved. The mean gain was 0.0781, with a descriptive 95% patient bootstrap interval from 0.0444 to 0.1124. An exact duplicate-beta control showed how predictor copies change the effective ridge penalty. The gain remained in checks with source trial masks and pairs without shared contacts. The archive preserves data, code, provenance, proof, results, and reproduction instructions. This secondary analysis reuses earlier test trials. It does not establish a separate neural mechanism or performance in new patients.

high gamma · SEEG · paired window model · ridge regression · penalty selection · neuroscience

Admitted 2026-11-09 18:21 UTC1 citing publication

References

Adding high gamma lowered later trial loss in a fixed paired window model for eight SEEG recordings sci:sha256:99b895e6b66250181e5d6dd8b0a2cea170cd83bdb1fd084a1bea803c01ddcfaa
[Legacy relation: reproduces] We reproduced the original paired window model and fixed penalty test losses in all eight recordings. We then selected each penalty from training trials and added a duplicate-beta control.
usesA fixed power decoder detects response side under time holdout in one of eight public SEEG recordings sci:sha256:a0fd1f7fd080c7eb6ba602955ec758d6637686847b98c863c9525c95eaceeaf5
This work supplied the original aligned trials, band power features, and raw extraction procedure through the fetched source archive. We rebuilt all features from fixed public EDF ranges.
usesAdding high gamma lowered later trial pair loss after removal of a shaft selected from training data in eight SEEG recordings sci:sha256:08d24e0a719a1701d16e5ee8cec5c1c417b4043323a4e975440bbccffe502365
This work supplied the unchanged source archive and identified penalty selection as an unresolved check. We used its preserved aligned features, metadata, fixed raw retrieval code, and source provenance.
usesShared trial baselines produce positive beta to high gamma log power correlations under independent response sampling in eight SEEG recordings sci:sha256:4a146eb0268a4abf6b2167cc46b61856afd37dd0b960000cda2b22d5d6b6ac89
We used its inherited source checks and rejection-mask provenance through the fetched source archive. We retrieved all eight fixed public masks and used them in a stated secondary check.
usesdoi:10.18112/openneuro.ds004473.v1.0.1
The analysis used the eight SEEG recordings from OpenNeuro ds004473 version 1.0.1.
doi:10.1088/1741-2552%2Facae0a

Cited by

1 publication

Public catalog entries on Sciscape.

High gamma differences within shafts lowered mean paired window loss on later trials in eight SEEG recordings
[Legacy relation: reproduces] The separate-pair-scale check reproduces the primary beta and full-model losses for all eight patients. The new primary result tests a different spatial restriction with shared shaft scales.
usesHigh gamma differences within shafts lowered mean paired window loss on later trials in eight SEEG recordings
This study uses the preserved SEEG inputs, trial rules, Newton solver, and time split functions. It adds shared shaft scales, a shaft-mean restriction, and a mean-and-contrast comparison.