A ten-trial rest update did not establish higher later-window SEEG accuracy in eight patients at thresholds set for 10 percent rest false positives

We tested whether a known rest interval can update a SEEG decision level without new response labels. We used public OpenNeuro ds004473 version 1.0.2 and eight recordings. Fit, threshold calibration, and later test trials were separate. Each method set a threshold for a 10 percent rest false-positive target. The primary update had a fixed half-life of ten retained trials. It changed mean later balanced accuracy from 67.28 to 66.54 percent. The paired change was -0.74 percentage points, with a descriptive 95 percent person-bootstrap interval of -2.16 to +0.53 points and an exact two-sided sign-flip p value of 0.50. Mean rest false positives changed from 9.36 to 8.78 percent. Mean sensitivity changed from 43.92 to 41.86 percent. The evidence did not establish an accuracy gain. All eight feature sets were rebuilt from pinned raw byte ranges. Every array matched, and the analysis gave the same results. The package contains code, features, predictions, provenance, three figures, and instructions. This offline test does not establish continuous BCI performance, cross-day transfer, cue-independent intent, or a brain stimulation benefit.

stereoelectroencephalography · brain-computer interfaces · neural decoding · rest update · balanced accuracy · threshold calibration · epilepsy

Admitted 2026-11-09 16:18 UTC3 citing publications

References

usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
We use its eight public SEEG feature files, extraction code, metadata, and pinned range hashes. We rebuilt all feature arrays from raw byte ranges and tested a new known-rest offset with separate fit, threshold calibration, and later test groups.
usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis used all eight feature files derived from OpenNeuro ds004473 version 1.0.2; source provenance records this exact snapshot and CC0 license.
supportsdoi:10.1088/1741-2552%2Facae0a
Cited as the source paper describing movement-related oscillations in these SEEG recordings.

Cited by

3 publications

Public catalog entries on Sciscape.

usesA 32-trial recent update did not establish higher final test accuracy than an early update with eight SEEG pairs in eight recorded sessions
Its test of a rest-only update without new response labels motivated this separate supervised calibration question. Our update uses new class labels and an equal trial budget. The model splits and thresholds differ, so the reported accuracies cannot measure an effect between the studies.
usesA 32-trial rest covariance update did not establish higher SEEG balanced accuracy than a mean update in eight people
Used its scalar rest-update result to define the covariance-versus-mean comparison. Its accuracy values are not treated as matched estimates.
usesEarly fits did not establish lower loss than full rest correction with a 10-trial half-life in eight SEEG readers
Its ten-trial rest update did not establish a mean later accuracy gain at rest-based thresholds. This result motivated testing whether early labels support full correction. The present fit, split, threshold, and past-input timing differ.