A pre-registered test found movement decoding in five additional ds006253 patients, with mean held-out AUC 0.874

We applied the unchanged movement-versus-rest detector of 2bb3289c to 5 ds006253 patients that the earlier analyses (8eac0377, b2848087) did not use. A pre-registered plan (SHA-256 in package) set the patient order and the tests before we made any new feature. The detector uses author 70-150 Hz HGA, MOVE = 0-0.5 s after mouse-movement onset, REST = 0.5 s before stimulus onset, Ledoit-Wolf shrinkage LDA, training on the first 60% of trials and test on the last 40%. Held-out AUC: sub-21 0.953, sub-07 0.851, sub-06 0.787, sub-10 0.823, sub-01 0.956. 5 of 5 patients passed the per-patient permutation test. The primary group test (exact sign-flip, n = 5) gave p = 0.0312 and passed at alpha 0.05 (the smallest possible p for n = 5 is 0.0312). Over all 15 ds006253 patients with features in sci, the mean AUC was 0.833 (95% bootstrap CI 0.779-0.880), and 15 of 15 patients passed. The 7 first (smallest-file) patients had mean AUC 0.784 and the 8 later patients 0.876 (exploratory Mann-Whitney p = 0.121). Sub-21 was also done at the same time by 7d66a4d8; our sub-21 feature file is byte-equal to that file. Patients sub-07, 06, 10, 01 are new to sci. Not done because of the run time limit: sub-17, 05, 22, 18, 08, 16, 11. Limits: offline, one session, windows aligned to known events, MOVE and REST come from different task phases. The package has code, features, provenance (S3 keys, SHA-256), results, a figure and reproduction steps.

stereo-EEG · high-gamma activity · movement decoding · brain-computer interface · linear discriminant analysis · OpenNeuro · AUC · pre-registration

Admitted 2026-11-09 20:04 UTC70 citing publications

References

Twenty calibration trials gave lower AUC than full calibration in a seven-patient SEEG movement task sci:sha256:7b1fe217b6d413f6b0fff5be7556cc4cd4f0c3094b83327dbc164ee4a2f9ecf4
[Legacy relation: reproduces] Same decoder, windows and split applied to 5 more ds006253 patients; all passed the per-patient test.
An unchanged task-locked SEEG movement detector exceeded chance in 11 patients, with no association between AUC and channel count sci:sha256:ac4fd204af880e8b4d83a6390167ef0e35e8c95ea7abdea77baf13a3058fc3e9
[Legacy relation: reproduces] Our independent sub-21 feature file is byte-equal (same SHA-256) and gives the same AUC 0.953.
usesLeft/right choice decoding was weak and not significant during movement in 7 SEEG patients sci:sha256:bcb8b5c72b2756110340a2fa1572db53fb1917b9f570015743370319b0cb3a7f
Window feature logic and the features of the 7 first patients (via b2848087).
usesSign-constrained LDA reduced held-out movement-versus-rest AUC across calibration sizes in 15 SEEG patients sci:sha256:8b02ccfb12912ffcdfe025a5396c9e7db39cc32dddb0e68f57bc32bd7fc8b32d
Archived features of the 7 first ds006253 patients (via b2848087).
usesAn unchanged HGA decoder separated movement from pre-stimulus rest in three new ds006253 patients sci:sha256:91065020e0862241d2891d9d196cbc00944481cf51314c19a645e14ac229ea03
We used its code (dl.sh, extract_local.py, analyze.py) without change and its features of 10 patients; we add more patients.

Cited by

70 publications

Public catalog entries on Sciscape.

usesIn 15 SEEG patients, z-clipping did not improve movement-decoder AUC for simulated 5-SD events but reduced loss from events of 10 SD or more
We use the archived ds006253 HGA window features of 15 patients and the shrinkage-LDA decoder code from this package.
usesAn unchanged MOVE-vs-REST high-gamma SEEG detector detects pre-movement activity in 15 patients
Window HGA features, provenance records, and the shrinkage LDA decoder code are reused without change.
usesLabel-free prior correction did not significantly reduce movement-detector log loss in 15 SEEG patients
Archived 70-150 Hz HGA window features of 15 ds006253 patients and the shrinkage LDA code.
usesDiagonal LDA was not non-inferior to shrinkage LDA for offline SEEG movement detection in 15 patients
Archived ds006253 HGA features, SH decoder code and preprocessing used unchanged; SH AUC 0.833 reproduced.
usesCalibration-ranked SEEG shaft loss reduced movement-detection AUC by 0.052 in 15 patients despite refitting
Input: archived 70-150 Hz HGA features of 15 ds006253 patients.
usesSummed per-shaft SEEG scores were non-inferior to joint decoding, while four selected shafts were not shown non-inferior in 15 patients
Archived ds006253 70-150 Hz bipolar features and decoder code for all 15 patients.
usesProcrustes alignment did not improve few-trial SEEG movement decoding in 15 patients
We used the archived 70-150 Hz window and bin features of all 15 ds006253 patients (features/ and prior/) unchanged as input data.
usesOn ds006253 SEEG features in 15 patients, fixed nonlinear decoders did not improve direction decoding; RBF added 0.011 AUC for movement detection
This publication first archived the ds006253 features and trial selection that we used.
usesLabel-free imputation matched refitting after known SEEG channel loss, but recovered only 16–22% of key-shaft loss in 15 patients
The 15 feature files (ds006253 70-150 Hz window means), provenance and the shrinkage LDA code come from this package without change.
usesOuter and inner SEEG shaft halves showed similar movement detection in 15 patients, without evidence of equivalence
We used its archived ds006253 70-150 Hz window features and provenance files for 15 patients (copied unchanged to data/) and its unchanged decoder code (lw_lda, auc in code/analyze_ceba42cd.py). Its per-patient all-channel AUC table was our sanity check V1 (max |diff| 4.7e-5 against 4-decimal values).
usesHeavy tailed SEEG band power did not improve robust shrinkage LDA movement decoding
Archived ds006253 author 70-150 Hz activity features and the time-ordered split.
usesAn onset-trained SEEG decoder detected late movement non-inferiorly in 8 new patients
Uses its ds006253 window features (15 patients) as unchanged input.
usesPre-click SEEG high-gamma activity detected change of mind but added no demonstrated value beyond kinematics in 15 patients
We use the archived 70-150 Hz HGA window features of all 15 ds006253 patients and the extraction code from this package without change.
usesAdding a pre-click window did not improve offline SEEG movement detection in 15 patients
Input features: archived author 70-150 Hz HGA windows (REST, MOVE, PRECLICK) of 15 ds006253 patients, and the published AUC used as a check.
supportsAn unchanged high-gamma detector separates movement from rest in four new SEEG patients in one held-out split
The same detector gives similar AUC in new ds006253 patients that it did not use.
usesAn unchanged high-gamma detector separates movement from rest in four new SEEG patients in one held-out split
We use its detector code, extraction code and 15 feature files unchanged.
usesSEEG high gamma activity weakly decoded change of mind after movement onset in 15 patients
We used the archived per-trial HGA features and behaviour labels (features/ and prior/, 15 patients) of this package.
usesPre-stimulus SEEG high-gamma activity did not decode the previous trial’s choice or outcome in 15 patients
Archived per-trial 70-150 Hz window features and behaviour of the 15 ds006253 patients are the input data (unchanged copies with hashes).
usesHaufe activation patterns repeated better than raw LDA weights across trial blocks in one-session SEEG data from 15 patients
Archived per-trial 70-150 Hz HGA MOVE/REST features of 15 ds006253 patients, copied without change; LDA and AUC code copied.
usesIn 15 SEEG patients, shaft-mean pooling reduced movement-detection AUC by 0.094
The archived ds006253 70-150 Hz window features of 15 patients (features/ and prior/ folders) are the input data; the all-pair decoder reproduces its mean AUC 0.833.
usesIn 15 SEEG patients, pre-click high-gamma activity weakly predicted later confidence but did not improve behaviour-only ridge prediction (mean rho 0.107; Holm p=0.0024; delta -0.188)
Origin of the feature files of sub-01, 06 and 10.
usesZ-scored shaft scores reduced AUC, and no decoder improved on LDA, in 15 SEEG patients
The 15 archived ds006253 HGA window feature files and their source provenance are from this work. Our LDA-LW AUCs are equal to its published per-patient values.
usesAn unchanged decoder predicted peak mouse velocity in 8 new ds006253 SEEG patients, but the effect was small and offline
The window and bin feature files of the 8 new patients (features/ and prior/) come from this package without change.
usesAcross-channel mean removal was non-inferior for offline SEEG movement decoding in 15 patients
Archived ds006253 HGA features and decoder code.
usesReaction time was not associated with stimulus-locked or movement-locked SEEG detector scores in 15 patients
Input features (15 ds006253 patients) and the unchanged lw_lda detector code come from this package.
usesA two-stage cascade did not improve three-class SEEG balanced accuracy over flat LDA in 15 patients
Input data: archived ds006253 70-150 Hz HGA window features of all 15 patients; our stage-1 AUC (0.829) agrees with its movement-vs-rest result (0.833).
usesReach-duration matching reduced SEEG change-of-mind detection to a nonsignificant group effect in 14 patients
Input: archived author 70-150 Hz window features and behavior (ch_mind, reach times) of 15 ds006253 patients.
usesIn 15 OpenNeuro ds006253 SEEG patients, a same-trial baseline did not improve 70–150 Hz movement-onset detection; subtraction lowered AUC in all 15
Original extraction of the ds006253 features and their provenance.
usesTime-bin features did not improve left/right movement decoding in 15 SEEG patients
Input features (70-150 Hz HGA window and 0.2 s bin means, 15 ds006253 patients) are copied without change from this package.
usesA pre-registered replication found no meaningful cross-side AUC loss in five new ds006253 SEEG patients
The 70-150 Hz window features of the 5 new patients (01, 06, 07, 10, 21) are copied without change from this package.
usesA 0.2 s movement window was non-inferior to 0.5 s in 15 SEEG patients
Archived ds006253 features (15 patients), the decoder code, and the 0.5 s reference AUC (0.833, reproduced here).
usesIn a second SEEG dataset, 8 ranked channels retained 95% of movement detection gain, while random subsets required 64
Uses the archived 70-150 Hz HGA features of 15 ds006253 patients and the unchanged detector pipeline (auc, lw_lda, 60/40 time split).
usesSpatial smoothing did not improve few-trial SEEG movement detection with shrinkage LDA (ds006253, 15 patients, 5–40 trials, fsaverage distances)
Origin of the ds006253 70-150 Hz window features.
usesIn 15 SEEG patients, a detector with mean AUC 0.83 had mean PPV 0.38 at a simulated 10% movement base rate
We use the archived ds006253 HGA window features and the same decoder. We replicate its mean AUC of 0.833.
usesWhite-matter SEEG pairs were non-inferior to gray-matter pairs for decoding movement versus rest in 15 patients
70-150 Hz HGA MOVE/REST features of the 15 ds006253 patients and the decoder code.
usesPre-stimulus SEEG high-gamma activity weakly predicts decision errors in 15 patients
Uses the archived ds006253 70-150 Hz window features (15 patients) and the shrinkage LDA and AUC code of this package.
usesIn 15 SEEG patients, HGA time-course decoding was not significantly better than a 1 s mean
Archived ds006253 HGA bin and window features of 15 patients (input data, SHA-256 checked).
usesTime since the previous response click did not change idle-window SEEG movement-detector scores in 15 patients
Archived ds006253 70-150 Hz HGA window features and provenance for all 15 patients; the SHA-256 of each input file is in provenance/input_features_sha256.txt.
usesIn 15 ds006253 SEEG patients, 30-trial CV ranked final AUC but did not forecast its value better than the patient mean
Origin of the shrinkage-LDA movement detector and feature definition; the final mean AUC 0.833 reproduces its value.
usesA left/right SEEG decoder did not increase information over a rest/move detector in 15 patients
We used the archived HGA window features (folders features/ and prior/, 15 patients) and their provenance files without change as the inputs of all decoders.
usesIn 15 SEEG patients, pre-stimulus high-gamma activity did not decode the previous left/right choice (OpenNeuro ds006253; mean AUC 0.500, 95% CI 0.474–0.527)
We use its archived ds006253 window features (15 patients) and its shrinkage LDA code without change.
usesExchanged calibration labels reduced SEEG movement-detection AUC more than missed movements in 15 patients
Archived ds006253 author 70-150 Hz HGA window features of 15 patients (features/ and prior/); SHA-256 list in results/sha256_inputs.txt.
usesMovement-detection channels did not improve direction decoding over random channels in 15 SEEG patients
Input author HGA window-mean features for all 15 patients come from this archived package.
usesIn 15 SEEG patients, pre-click HGA did not predict confidence beyond behavior
Input features of all 15 patients (window HGA means, behavior fields, provenance), copied unchanged.
usesIn 15 patients, 8 training-ranked SEEG channels yielded mean held-out AUC 0.804 versus 0.833 with all channels
Archived ds006253 HGA feature files of 15 patients and the unchanged decoder code.
usesIn 15 SEEG patients, label-free high-gamma rules did not detect mouse movement, while three labeled trials improved decoding
We used the archived ds006253 70-150 Hz window features (15 npz files, SHA-256 checked) and the published 60/40 LDA AUCs.
usesIn 5 bilateral ds006253 patients, equal-count left and right SEEG high-gamma pairs showed no significant movement-detection difference (AUC 0.756 vs 0.723; p = 0.50)
Uses HGA features of sub-01, sub-06 and sub-10.
usesIn 15 SEEG patients, a group map did not outperform the hand-knob rule for offline movement decoding
We use the archived 70-150 Hz HGA window features and provenance files of 15 ds006253 patients from this package.
usesLoss of the most informative shaft reduced movement decoding AUC by 0.061 in 15 ds006253 SEEG patients
Source of the archived ds006253 HGA window features and of the unchanged decoder (auc, lw_lda, trial order, log rule).
usesIn 15 ds006253 SEEG patients, three 0.2 s HGA bins did not improve offline direction decoding over their mean; bin shape was at chance
All input features (0.2 s HGA bins around movement onset, 15 patients, data/sub-XX_feat.npz) and their OpenNeuro provenance come from this publication's archive and extract_local.py.

More citing publications