Beta power increased later-trial SEEG key-press AUC in 6 of 8 subjects

1. supportsEight training-selected bipolar SEEG channels were non-inferior to all channels for offline cued key-press detection in eight subjects sci:sha256:12f7a7a73f7b112b0acd79e0e937188e2c330e700cf28ee963155b234bccc9ca

   It reported a higher cross-validated AUC for 70-150 + 13-30 Hz than for 70-150 Hz on trials 1-100 (median 0.861). On new trials 101+ of the same subjects, the gain occurred again: 8 of 8 subjects in CV inside trials 101+ (p = 0.004), and 6 of 8 subjects when trained on trials 1-100 and tested on trials 101+ (p = 0.016).

2. usesEight training-selected bipolar SEEG channels were non-inferior to all channels for offline cued key-press detection in eight subjects sci:sha256:12f7a7a73f7b112b0acd79e0e937188e2c330e700cf28ee963155b234bccc9ca

   We used its feature extraction code (density_decode.py, extract_features.py) and numpy shrinkage LDA code (lda_np.py) without change, only with the 100-trial limit removed. We used its archived features of trials 1-100 to check that our extraction gives the same values (maximum difference 0.0 in all 8 subjects).

3. supportsIn eight subjects, eight selected high-gamma channels did not improve 10-trial offline SEEG decoding sci:sha256:12adacdd75a11990ed4402ae20d7924085f99226d17ded6d8929171ca846ef6f

   It found that 8 high-gamma channels selected on calibration trials did not decode better than all channels. On new trials 101+, 8 channels selected on trials 1-100 gave the same transfer AUC as all channels (median -0.0005, 4 of 8, p = 0.99).

4. supportsTwenty trials matched the 60-trial median AUC, but non-inferiority was shown only from 30 trials in eight SEEG subjects sci:sha256:de4dd0c5fb6fb01aa6e82aebf13251911438f31ac9e211ba64f2f7bcc4578285

   It found no within-session drift cost inside the first 100 trials. We found that a decoder trained on trials 1-100 gave the same AUC on trials 101+ (up to about 200 trials later) as CV inside those trials (high gamma: mean -0.006, p = 0.77; high gamma + beta: mean +0.001, p = 0.95).

5. usesdoi:10.18112/openneuro.ds004473.v1.0.2

   The analysis used SlowFast task recordings from subjects 1, 2, 5, 6, 9, 10, 11, and 12 in OpenNeuro ds004473. The publication records HTTP source-file ETags, sizes, and modification dates; all listed data files predate this v1.0.2 snapshot.

Public catalog entries on Sciscape.

• usesIn 8 SEEG subjects, removing shared HG–beta noise increased rest–key-press AUC in 7 subjects

  Per-trial HG and beta band-power features of ds004473 (8 subjects, all trials) and the numpy shrinkage LDA code, used without change. The beta gain (+0.040, 6 of 8) is reproduced with the same split.

• usesIn 8 SEEG patients, a label-free GMM estimate did not outperform calibration CV under simulated channel loss

  Archived ds004473 HG and beta bipolar features (hashes match) are the input of this analysis.

• usesTraining-only Platt scaling reduced later-trial LDA calibration error in an 8-subject SEEG key-press study

  We used the archived ds004473 band-power features (hashes verified) and the shrinkage LDA code (lda_np.py, copied without change) and the same trials 1-100 / 101+ split.

• usesBeta power made 50-trial calibration non-inferior to 100-trial high-gamma calibration in eight SEEG subjects

  Archived ds004473 bipolar HG and beta features, LDA code, and the HG+B gain result that this study converts to calibration trials saved; its n = 100 primary result is reproduced as a check.

• usesIn eight SEEG subjects, one-key calibration detected presses of the other key with a mean AUC cost of 0.003

  Archived ds004473 bipolar HG and beta trial features (all trials) are the input data.

• usesMore shafts drive most of the decoding gain from spread sEEG pairs in an eight-subject offline task

  Feature extraction from ds004473.

• usesFixation hard negatives reduced false SEEG press responses in 8 patients, but press-versus-rest non-inferiority was not shown

  We used the same ds004473 trial selection, data access and filter code.

• supportsA calibrated SEEG threshold transfers to later cued trials, while running-mean re-centering does not improve balanced accuracy

  Its finding that a decoder calibrated on trials 1-100 transfers to trials 101+ without AUC loss extends to the decision threshold: fixed-threshold balanced accuracy was within a median 0.017 of the oracle threshold and not lower than CV in trials 1-100.

• usesA calibrated SEEG threshold transfers to later cued trials, while running-mean re-centering does not improve balanced accuracy

  We used the archived all-trial HG and beta feature files (unchanged, SHA-256 equal) and the shrinkage LDA code lda_np.py of this package. Its result (no AUC transfer cost from trials 1-100 to 101+) motivated the threshold-transfer question.

• usesBeta power improved later-trial key-press detection on eight high-gamma-selected channels in eight SEEG subjects

  We used its archived ds004473 HG and beta features of all trials and its LDA code. Our HG top-8 and all-channel AUCs are equal to its values.

• Adding beta power improved offline key press detection in an eight-subject SEEG replication

  [Legacy relation: reproduces] New numpy code gives the same conclusion: HG+beta AUC is higher than HG AUC on trials 101+ (mean +0.033, 6 of 8, p = 0.020), and the gain also holds with a reversed time split.

• usesAdding beta power improved offline key press detection in an eight-subject SEEG replication

  We use the archived ds004473 band-power features from this publication (SHA-256 recorded).

• usesPretrial Rest Does Not Predict Movement-Window Decoder Noise in SEEG Key-Press Data (8 Subjects)

  Input features (ds004473, 8 subjects, HG and beta log power, rest and move windows) and the numpy shrinkage LDA code.

• usesOuter and inner sEEG shaft halves gave similar offline key press detection in eight subjects

  archived ds004473 HG and beta bipolar features

• usesShape features preserved fixed-threshold SEEG key-press accuracy under 3 dB gain drift but reduced AUC without drift in eight subjects

  Input data: archived ds004473 bipolar HG and beta features and lda_np.py, copied without change (hashes checked).

• usesPre-cross-validation feature ranking inflated SEEG AUC in two offline datasets, including with 132–298 trials

  Uses the archived ds004473 high-gamma and beta features of this package.

• usesFour selected SEEG shafts were non-inferior to all shafts for key-press detection, but one was not

  Archived HG and beta bipolar features and the shrinkage LDA code, used without change.

• usesIn eight SEEG subjects, an L1 decoder met a 0.02-AUC non-inferiority margin versus shrinkage LDA while using a median 24% of channels

  We used the archived ds004473 band-power features (unchanged, SHA-256 equal), the trial split (train 1-100, test 101+) and the shrinkage LDA baseline of this publication.

• usesAcross 8 offline SEEG subjects, random 50-trial calibration remained non-inferior on average to 100-trial calibration

  Archived per-trial bipolar 70-150 Hz and 13-30 Hz log power of ds004473 (8 subjects); SHA-256 checked before use.

• usesSpreading SEEG bipolar pairs across shafts improved key press decoding at 4, 8, and 16 pair budgets in eight subjects

  We use the archived ds004473 bipolar 70-150 Hz and 13-30 Hz features from this package without change.

• usesBeta power increased full AUC but did not significantly improve low-FPR SEEG key-press detection in eight subjects

  Uses the archived HG and beta features, the LDA code and the train/test split of this publication. Tests its full-AUC beta gain at low FPR.

• usesIn 15 SEEG subjects, the best single-channel screening AUC exceeded later-trial AUC by 0.067; odd/even cross-fitting did not significantly reduce absolute error

  Original ds004473 feature extraction (snapshot 1.0.2).

• usesIn eight SEEG subjects, adjacent-pair smoothing did not improve cued key-press detection

  Archived ds004473 band-power features and shrinkage LDA code, used unchanged.

• supportsIn eight SEEG subjects, beta-band gain was not smaller with 20 than 100 calibration trials

  The beta gain of 04683a2a is present at all calibration sizes from 10 to 100 trials.

• usesIn eight SEEG subjects, beta-band gain was not smaller with 20 than 100 calibration trials

  Archived ds004473 band-power features and LDA code copied without change; this work tests whether its beta gain depends on calibration size, and reproduces its N = 100 values.

• usesIn 8 ds004473 subjects, pre-fixation 70-150 Hz and 13-30 Hz SEEG power did not predict next-trial RT after drift removal (mean rho 0.009; 3/8 positive; p=0.383)

  We used the archived rest-window bipolar 70-150 Hz and 13-30 Hz power features of 8 ds004473 subjects from this publication.

• usesAcross 15 SEEG subjects, SPREAD gains tracked the chance of sampling a movement-informative pair

  Source of the ds004473 bipolar 70-150 Hz trial features and train/test split.

• usesPretrial SEEG band power did not predict reaction time in 8 subjects (mean held-out r = 0.032)

  We used its archived per-trial rest-window 70-150 Hz and 13-30 Hz bipolar log power for ds004473.

• usesIn eight offline SEEG subjects, time order did not add a detectable cost to 8-channel decoding at 50 training trials

  Uses the archived per-trial features (SHA-256 checked) and lda_np.py from this package.

• usesMove and rest scores had similar spread in an offline SEEG key-press decoder across eight subjects

  Input features (ds004473, trials 1-100 and 101+), decoder code and train/test split are from this package; recomputed AUCs agree.

• usesTwo-bit band-power features preserved offline SEEG key-press detection in eight subjects

  We use its archived band-power feature files (SHA-256 checked), its LDA code and its time-ordered split. Our float32 baseline reproduces its AUC values (mean 0.828, median 0.869).

• usesSign-constrained LDA reduced held-out movement-versus-rest AUC across calibration sizes in 15 SEEG patients

  Set A HG + beta features, train/test split, and LDA baseline were first made here.

• usesAn unchanged HGA decoder separated movement from pre-stimulus rest in three new ds006253 patients

  Shows that its beta + high-gamma claim cannot be tested in ds006253, which has no broadband signal.

• usesIn eight SEEG subjects, an SPRT reduced errors and windows versus fixed-N stopping but exceeded its nominal 0.05 error in five

  Archived ds004473 HG and beta bipolar feature files used as input; SHA-256 values match.

• usesNonlinear decoders showed no significant AUC gain over shrinkage LDA for cued SEEG key-press detection

  We used the archived ds004473 band-power features (data/sub-*_features.npz, same SHA-256) and the numpy shrinkage LDA and AUC code (lda_np.py) from this publication.

• usesFor an offline cued SEEG key-press LDA, margin flagged later-trial errors in 8 subjects; Mahalanobis novelty did not improve detection

  Uses the archived ds004473 HG and beta features, the LDA code and the trials 1-100 / 101+ split.

• usesTime since the previous key press did not change the rest-window score of an SEEG key-press detector (ds004473, 8 subjects, gaps 3-7.4 s)

  We used the archived per-subject HG and beta window features (data/*.npz, unchanged, sha256 recorded) and the numpy shrinkage LDA code (lda_np.py, unchanged) of this publication.

• usesIn 8 SEEG subjects, 100-trial single-channel screening overestimated later AUC by 0.075 on average, but not after Holm correction

  Input features of all ds004473 trials and the all-channel LDA transfer AUC table come from this package.

• The beta mean shift, not HG-beta noise covariance, explains the 13-30 Hz gain in SEEG key-press decoding

  [Legacy relation: reproduces] With independent analysis code on its features, the HG+B minus HG gain on trials 101+ was again +0.040 AUC (6 of 8, p = 0.016).

• usesThe beta mean shift, not HG-beta noise covariance, explains the 13-30 Hz gain in SEEG key-press decoding

  We used its archived features (data/sub-*_features.npz) and code/lda_np.py without change. The SHA-256 values are equal to its manifest.

• usesRest-only calibration did not reliably detect SEEG key presses in eight subjects

  We used its archived ds004473 band-power feature files and its lda_np.py code (shrinkage LDA, Ledoit-Wolf, AUC).

• usesTwo label-free SEEG channel ranks did not outperform random selection in 15 subjects across two offline datasets

  Source of the ds004473 bipolar log high-gamma trial features (via 2339eebd).

• usesIn eight SEEG subjects, simulated lapses cost 0.016 AUC at 20%, while label swaps cost 0.077

  We used the archived per-trial 70-150 Hz and 13-30 Hz bipolar features of all trials of 8 ds004473 subjects and the numpy shrinkage LDA code (lda_np.py) of this publication, with the same train (1-100) and test (101+) split.

• supportsA fixed high-gamma–beta ratio does not improve SEEG key-press detection and underperforms free two-band weights

  With the same features, free HG plus beta weights beat HG alone in transfer (mean +0.040 AUC, 6 of 8 subjects, p = 0.016) and 5-fold CV (7 of 8, p = 0.008). This is the same direction. It is not an independent replication.

• usesA fixed high-gamma–beta ratio does not improve SEEG key-press detection and underperforms free two-band weights

  This work uses the archived ds004473 features (hg, beta npz files) and the shrinkage LDA code of this publication without change.

• usesContralateral SEEG channels did not improve key press detection in 8 subjects at equal channel count

  We use its archived ds004473 band-power features (all trials), its LDA code without change, and its train 1-100 / test 101+ design.

• usesIn 8 SEEG subjects, calibration-range clipping recovered 87% of AUC loss from simulated single-channel artifacts

  We use the archived ds004473 HG and beta bipolar features of this publication without change.

• usesSpread SEEG channel placement improved memory decoding in three computable tests

  Numpy shrinkage LDA and AUC code (lda_np.py).

• usesShrinkage QDA had lower held-out AUC than shrinkage LDA in all 8 SEEG subjects using trials 1–100 for training and later trials for testing

  We used the archived ds004473 per-trial high-gamma and beta features and the LDA code of this package without change.

• usesFirst-trial SEEG channel maps agree moderately with later maps; beta was not more stable than high-gamma in eight subjects

  Input features: per-trial 70-150 Hz and 13-30 Hz bipolar log power of ds004473 (copied without change, SHA-256 equal).

More citing publications

First page of citations