Zero sample runs continued after all 932 marked stimulation intervals in 250 selected hippocampal SEEG records
We tested exact constant sample runs in 250 selected hippocampal SEEG records from seven people in OpenNeuro ds007095 version 1.0.0. Both channels had a zero run at all 932 marked stimulation intervals. Runs lasted 215 or 325 ms. Their exclusive ends were 115 or 225 ms after the marked end. Source interval masks left 65.14% of these zero samples outside the masks. No native sample reached either declared digital limit. In the planned test, one of 250 windows from 5 to 15 s after the last marked end had a long constant run. That run was at the file end. No before-stimulation window had such a run. The archive supplies native inputs, source versions and hashes, code, complete result tables, three figures, and offline quality masks. A separate reader matched all 9.15 million sample codes. A fresh run reproduced all 14 numerical result files. The findings concern saved codes in this selection. They do not establish hardware cause or neural recovery time.
stereoelectroencephalography · SEEG · responsive neurostimulation · sample quality · constant runs · OpenNeuro · hippocampus
References
4- usesBilateral ridge forecasts gave no clear 50 ms error reduction across later sessions in seven people at equal input count sci:sha256:4c59af20541983b231c1367be8f2365ac281d285fadea1c78ad7fc4d754db3b5
We reused its fixed 250-record hippocampal depth selection, complete native EDF files and sidecars, source receipts, full selection ledger, EDF reader, and retrieval code. This reuse enabled a new native-code quality audit. It does not add a new participant cohort.
- usesConstant SEEG runs continue after nominal stimulation ends in two ds005558 recordings from one person sci:sha256:78861a25c9795deb6c7b36d69534cf13a8cfed5003b2120cc8496832097f6d94
Its exact native-code run method and declared digital-limit check supplied a method precedent. We adapted the checks to a different public dataset with seven people, two hippocampal channels, and 200 Hz sampling. We did not reuse its signal data.
- usesdoi:10.18112/openneuro.ds007095.v1.0.0
Source data analyzed in this publication: OpenNeuro ds007095 version 1.0.0, RNS_Epilepsy-iBIDS; the saved dataset metadata identifies this exact version and CC0 license.
- doi:10.3389/fnins.2026.1815732
Cited by
5 publicationsPublic catalog entries on Sciscape.
- usesAn exact check gave more 7 Hz phase certificates after SEEG gaps in 250 records with bounded missing voltage
Used all 250 selected source records, channel and event sidecars, the selection ledger, source receipts, the EDF reader, and retrieval code. Rechecked saved zero runs and tested conditional phase limits after the last stimulation gap.
- usesTwenty-second calibration in the current record gave no clear theta power gain over calibration from the previous session in seven people with depth SEEG
We used its fixed 250-record depth selection, native EDF files and sidecars, pinned source versions and hashes, complete selection ledger, EDF reader, and retrieval utility. These inputs enabled a new chronological test of theta band calibration. We downloaded all 250 native EDF files again and checked their hashes. The reused people do not form a new cohort.
- usesA fixed 50 ms phase forecast had 73.87 degrees more error with a centered target in 250 SEEG records at 7 Hz with a 1 s Hann filter
This publication defined the inherited record timing rules and source selection ledger. Its rules permitted tests on intact voltage before stimulation.
- usesTraining on six other people reduced unseen contribution error by 4.94% in 250 SEEG records. The centered target used a 7 Hz, 1 s filter at a 50 ms delay.
This work supplied the inherited record timing rules and source selection ledger. Those rules permit the early fits and later reference targets to end before marked stimulation.
- usesSource interval checks accepted theta features from one-second windows with saved zeros in 250 selected SEEG records
We used its fixed 250-record selection, complete native SEEG files, pinned metadata, and reader code. Its zero-run evidence set the source gap lengths. We added feature validity counts, a controlled power test, and a finite feature mask.