A 4 to 30 Hz voltage veto reduced high-gamma crossing excess near discharge labels in 23 SEEG sleep records

We tested a voltage rule for a high-gamma reader in public SEEG sleep data. The rule blocks a 70 to 110 Hz crossing when the recent 4 to 30 Hz peak exceeds its limit. The test used 23 people and 425 expert discharge labels. Mean crossing excess near labels fell from 22.01 to 4.97 percentage points. The reduction was 17.05 points, with a 95% bootstrap interval of 11.28 to 23.29 points. A shifted rule reduced excess by 1.15 points. The rule lowered detection of controlled 300 ms 90 Hz signals by 0.51 points. With added 15 Hz activity, the loss was 2.99 points. The tests used sleep data and artificial signals. BCI task decoding and brain stimulation remain untested. The package preserves features, source hashes, code, figures, trial results, and a source retrieval recipe.

stereoelectroencephalography · high-gamma reader · voltage veto · interictal discharges · brain-computer interface · signal detection

Admitted 2026-11-09 16:44 UTC2 citing publications

References

usesAdjacent contact subtraction did not consistently reduce 70 to 110 Hz power threshold crossings near labeled discharges in 22 SEEG sleep records sci:sha256:156fdfdba863d38f7bc29e0f44f8f0c1a19a997d8383184ab1d0635a2d148500
The EDF reader, contact rules, source descriptions, and source hashes were used. All source files were checked again. The prior power arrays were used to verify new extraction.
A 500 ms median reduced excess 70 to 110 Hz power threshold crossings near labeled discharges in 23 SEEG sleep records sci:sha256:e2ef6d5197b34a330862029e345bcb3ac89364ca6c2a9bd7b87c887c64f61a05
[Legacy relation: reproduces] The event and background fractions for the five-block mean and median readers were reproduced. The new voltage veto and controlled voltage additions extend this comparison.
usesdoi:10.6084/m9.figshare.26131978.v3
The study analyzed the released Figshare version 3 SEEG sleep dataset.
doi:10.1038/s41597-024-04187-y

Cited by

2 publications

Public catalog entries on Sciscape.

Unreviewed veto calibration raised crossing excess by 8.45 points in 23 SEEG sleep records with a fixed reader
[Legacy relation: reproduces] The ungated and reviewed 99th-percentile veto reference calculations reproduced the parent event fractions, background fractions, crossing excess, 23-person count, and 425-label count exactly. The new study uses a different bootstrap seed and does not claim an independent cohort replication.
usesUnreviewed veto calibration raised crossing excess by 8.45 points in 23 SEEG sleep records with a fixed reader
We used the MIT-licensed EDF reader and feature extraction code, fixed adjacent-contact descriptions, source member hashes, compact high-gamma and low-frequency feature arrays, and controlled voltage test rules. We retrieved the original source archive and checked all 216 member hashes. A new extraction from all 25 EDF files matched all 100 feature arrays. These inputs enabled the new comparison of reviewed, empirical unreviewed, and robust unreviewed veto calibration.
usesA 200 ms delay in a fixed voltage veto raised high-gamma crossing excess by 4.28 percentage points in 23 SEEG sleep records
Uses all 25 compact feature records and contact descriptions, the causal filters and fixed thresholds, the EDF reader and retrieval code, and source hashes. Reproduces the no-veto, current-veto, and shifted-control scores exactly. Tests new warning delays on the same cohort.