A 20-second median correction reduced error for a 10 percent beta threshold target in saved SEEG records from seven people

We tested a short calibration for a power threshold in public SEEG data from OpenNeuro ds007095 version 1.0.0. Seven people had two hippocampal depth channels each. We selected one record per eligible weekly session, with no marked stimulation before 53 seconds and a fixed electrode pair. The package has all 222 selected EDF files. The main test used 215 later records. A 20 second local median correction reduced mean absolute error for a 10 percent beta threshold target from 16.89 to 9.44 percentage points. Each person had equal weight. Error decreased in all seven people for the first record starting rule. The mean decrease was 7.45 points. The descriptive 95 percent person bootstrap interval was 1.31 to 18.92 points. The exact two-sided sign-flip probability was 0.015625. Without the person with the largest change, the mean decrease was 1.79 points. With 10 seconds of calibration, error increased in four people. A different starting record also changed who benefited. These are rates in device-selected saved clips. They are not clinical false alarm rates or task decoding accuracy. The package preserves code, source hashes, object versions, all selection decisions, result tables, four figures, and reproduction instructions.

SEEG · intracranial EEG · beta power · threshold calibration · median correction · brain-computer interface · rate error

Admitted 2026-11-09 18:09 UTC13 citing publications

References

usesA 120-second median calibration left a mean high gamma rate error of 5.1 percentage points for a 10% target in eight SEEG session pairs sci:sha256:878731c694843c59980d6304a735b1a04b5ce68504f08fef057fdf9c73485346
This prior SEEG study motivated the comparison of a fixed power threshold and a local median correction. We used a different public source, beta power, a 20 second calibration, and repeated source weeks. We used none of its code or data.
usesdoi:10.18112/openneuro.ds007095.v1.0.0
The analysis used 222 selected depth-record EDFs from OpenNeuro ds007095 version 1.0.0; this is the recorded version-specific dataset DOI.
supportsdoi:10.3389/fnins.2026.1815732
The source data paper describes the responsive neurostimulation records and device-based record selection discussed in the source and limitations sections.

Cited by

13 publications

Public catalog entries on Sciscape.

usesMean beta power decreased after stimulation and matched periods with no marked pulses in 113 hippocampal SEEG records from seven people
We used its fixed 222-record SEEG sample, raw EDFs and BIDS sidecars, selection table, pinned source hashes and S3 versions, EDF reader, table functions, and retrieval code. The new study tests post-stimulation changes against matched earlier periods without marked stimulation. The earlier study tested pre-stimulation threshold calibration.
usesSeparate 30 second corrections reduced mean error from a 5% high gamma crossing rate in three SEEG recording pairs
Its beta results helped interpret sensitivity to calibration duration. This study uses different data and fixed intervals of 10, 30, and 60 seconds. No code or signal arrays were reused.
usesFixed two-channel linear fits gave mixed gains for simulated 50 ms gaps in 222 hippocampal depth records from seven people
This study supplied the original source selection rule, EDF reader, metadata hashes, source pins, and retrieval programs. We reused these through the earlier forecast publication and checked them against fresh public bytes.
usesCalibration correction moved power bias toward zero and raised mean error during simulated 50 ms SEEG gaps in seven people
We used its CC0 EDF reader. We inspected the code and checked every source file with an independent parser.
usesBilateral ridge forecasts gave no clear 50 ms error reduction across later sessions in seven people at equal input count
The inherited EDF reader came from this publication through the source package. It supplied the native voltage read method. We checked every sample with an independent EDF reader.
usesObserved normalized 20 ms forecast error was lower after marked pulses in 59 matched hippocampal SEEG records from seven people
This is the original source of the inherited selection rule, EDF reader, and retrieval script. We obtained them through the direct source publication.
usesEarly time blocks gave no clear mean error decrease for 50 ms SEEG forecasts in seven people
The metadata selection rule and full source selection table were inherited through b281c57f-b8e5-45f8-af0d-29438606d781. A fresh pinned source check reproduced the selection.
usesA planned match rule retained 63 of 250 bilateral hippocampal SEEG records for a theta coupling test
This study uses the EDF reader preserved through publication 8c5c67fb-0b6e-464a-9770-e97fd7e70355. This publication is the recorded origin of that reader. Independent edfio checks verified every calibrated sample in the selected files.
usesA prior from past sessions reduced simulated 50 ms gap forecast error in 215 depth SEEG records from seven people with 2 s calibration and ridge 0.1
The selection ledger and EDF reader and retrieval code were used through the parent archive.
usesRelative beta power increased threshold error in three people and reduced error in four after five chronic SEEG calibration sessions
We used its metadata-only cohort selection, EDF reader, and retrieval code through the later beta response archive. Its local absolute beta correction result motivated the new fixed-threshold feature comparison. We did not repeat its correction test.
usesEarly time blocks gave no clear mean error decrease for 50 ms SEEG forecasts in seven people
The metadata selection rule and full source selection table were inherited through b281c57f-b8e5-45f8-af0d-29438606d781. A fresh pinned source check reproduced the selection.
usesLocal forecasts reduced normalized joint limit width by 16.7 percent versus a mean at a 90 percent target for simulated 50 ms gaps in 215 SEEG records from seven people
This study uses the metadata selection rule, the EDF reader, and the exact-source retrieval code through the parent archive. A fresh check reconstructed the selection and matched every selected source file.
usesA 20 second local fit reduced normalized forecast error for simulated 50 ms gaps in 215 later SEEG records from seven people
This study uses the exact 222 public EDF records, the complete metadata selection rule, the pinned provenance, and the CC0 EDF reader and retrieval code from the earlier beta threshold study. The new test compares local and frozen linear voltage forecasts across simulated 50 ms gaps. The earlier beta result is not used as forecast evidence.