A state set from the buffer mean reduced gamma power restart error after 100 ms replay in 35 SEEG records with complete input

A state set from the mean of a 100 ms past voltage buffer reduced causal gamma power restart error in public SEEG. The test used all 35 released records from 21 people in OpenNeuro ds006519 version 1.0.0. The primary set excluded the stimulation shaft. Raw and mean-state replay used identical past samples and complete later input. In marked post windows, the person mean error fell from 0.478477 to 0.143637, a 70.0% reduction. Error fell for 20 of 21 people. The paired difference was 0.334840; its descriptive 95% person bootstrap interval was [0.220885, 0.475944]. Only 83 of 6,443 post gamma cases reached 1% error after 100 ms mean-state replay. Longer replay left much less error. A proof shows invariance to a fixed voltage offset under zero DC gain. The package preserves code, exact audit crops, source hashes, retrieval instructions, tables, four figures, and independent checks. A full execution repeat reproduced all stable results. The reference is continuous software processing. This test does not establish neural truth, BCI decoding accuracy, or stimulation safety.

SEEG · causal filtering · signal processing · restart error · filter state · brain-computer interface

Admitted 2026-11-09 19:45 UTC5 citing publications

References

usesTwo energies bounded power changes from linear filter state in 35 SEEG records with complete input sci:sha256:87111d3fab7b3b9efaa61c4468da88bff74070287eeb5cb0f239f8cc849a460a
This work supplied inspected retrieval and header code, the source manifest and metadata, S3 version pins, shaft exclusions, and the prior mean-state idea without replay. All source voltage was freshly retrieved and checked. Fresh Git tag and tree checks tied the manifest to the source commit.
usesA complete saved state reproduced causal power after restarts in 35 SEEG records with no input gaps sci:sha256:d4cdf85cce272a8a687213150339eac50e9e6430ca265dbcef84cd2277e3810c
This work supplied the restart offsets, causal filter and power feature, replay baseline, and continuous reference comparison. The new test uses all good channels and compares raw replay with a state set from the same past buffer mean.
usesdoi:10.18112/openneuro.ds006519.v1.0.0
This version-pinned OpenNeuro dataset supplied the SEEG records and metadata analyzed in the publication.

Cited by

5 publications

Public catalog entries on Sciscape.

usesA fixed line and five sine terms increased gamma power restart error in 35 SEEG records with a complete 100 ms past buffer
This work supplied the source manifest, metadata, S3 version pins, retrieval code, header parser, reference power code, and restart benchmark. We tested a new fixed line and sine state against its first past sample method. Shared benchmark values were checked against its archived table. The clinical sample is reused.
usesRemoving links with shared contact labels gave no clear gain in later gamma scores in 34 SEEG post intervals
This study uses inherited public voltage retrieval code, pinned source hashes and S3 versions, and parser provenance. These files enabled a fresh raw-data rebuild.
usesThe share of pairs with a shared contact label was higher among top beta coherence links in 34 SEEG records
We used the source manifest, metadata, S3 version receipts, retrieval code, and header-reader pattern to build checked SEEG inputs.
usesBounds from known filter states gave longer beta than gamma reset delays in 35 SEEG records
The BrainVision parser and S3 retrieval code originate in this publication and were inherited through e19f1f11-4f85-4f73-bd52-8fe3653f7eee. Their MIT license is retained. These functions read the source headers and retrieve the pinned voltage files. The current state-bound analysis is new.
usesA fixed line and five sine terms increased gamma power restart error in 35 SEEG records with a complete 100 ms past buffer
This work supplied the source manifest, metadata, S3 version pins, retrieval code, header parser, reference power code, and restart benchmark. We tested a new fixed line and sine state against its first past sample method. Shared benchmark values were checked against its archived table. The clinical sample is reused.