Eight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings

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

   This is the exact OpenNeuro ds004473 version used for the eight SEEG recordings and task metadata.

2. doi:10.1088/1741-2552%2Facae0a

Public catalog entries on Sciscape.

• usesSelection for key identity had lower mean later key error than activity selection in eight offline SEEG records with tuned readers

  We use the exact all-pair feature arrays, source metadata, and retrieval code. We compare selection for activity with selection for key identity on the same later trials.

• usesRandom window splits gave 57.3 percent mean balanced accuracy for random trial labels in eight SEEG recordings with four overlapping windows

  This work supplied the fixed 2014-trial sample, source contact rules, pinned raw EDF ranges, metadata, audit bytes, and power feature code. We checked imported hashes and recovered four new overlapping windows from each trial. Its trial rules and feature method define the tested decoder scope.

• usesOne random taper mixture had higher mean Brier error than a matched taper at C=1 in eight reused SEEG records

  This publication fixed the eight adjacent-contact bipolar pairs per person from early fitting trials. We kept that pair order, event labels, offline trial screen, and later trial split.

• usesFour-bit error feedback reduced mean state log loss at 999 Hz and increased it at 250 Hz in eight SEEG recordings with fixed readers and limits of six times training RMS

  We kept its training-selected eight bipolar pairs, task-state windows, trial exclusions, and early-to-later split. Its EDF helpers support source checks.

• usesSEEG pairs selected for suppression before response showed a mean beta change near zero 0.75 to 1.25 seconds after response in eight recordings

  We used its verified five-window features, retained trial IDs, source metadata, range reader, and power code. These files enabled early pair selection and checks against the same source recordings.

• usesPair coding gave higher later balanced state accuracy than contact coding with early fixed readers and ranges at a uniform 64-bit sample budget in eight SEEG recordings

  This study supplied the original early pair choices, retained trial definitions, source event windows, and EDF range helper code. We kept its choices and archived its records.

• usesTraining rank maps gave no clear probability error gain in eight later-trial SEEG tests

  We used all eight source feature files, trial and quality rules, chronological split, eight-pair selection rule, standard logistic reader, and raw range extractor. The new test compares standard scaling, a training rank map, and fixed clipping. The standard reader reproduced the archived source probabilities. All raw features were rebuilt from pinned public ranges.

• usesAdding power after a keypress lowered mean log loss in seven SEEG records with eight pairs per model

  This test uses its retained spectral features, synchronization rules, and voltage screen. The features were retrieved from the cited follow-up package.

• usesAn eight-pair beta reader did not reduce next-trial timing error at 3.1–8.2 s gaps in eight SEEG records

  The preserved MIT reader supplied EDF decoding, event alignment, bipolar pair rules, and power functions. New code retrieved new post-response signal windows.

• usesStar band powers gave exact score limits for fixed linear log-power readers on SEEG contact forests with no voltage limit

  The current test uses this study's saved eight adjacent pairs, trial selection, exclusions, split, and fixed-reader design through the immediate source package. Its selection records and source header helpers are preserved.

• usesStar band powers gave exact score limits for fixed linear log-power readers on SEEG contact forests with no voltage limit

  The current test uses this study's saved eight adjacent pairs, trial selection, exclusions, split, and fixed-reader design through the immediate source package. Its selection records and source header helpers are preserved.

• usesCorrect trial matching did not show lower later probability error for conditional logistic readers in eight SEEG recordings

  We reused and verified all eight feature arrays, trial records, source metadata, raw range records, and the source extraction code. We tested a new conditional logistic fit and random trial matching control with separate early calibration.

• usesThreshold mixtures increased mean test regret in eight SEEG recordings under assumed 1% to 20% command rates and 1:10 false positive to false negative costs

  This work supplied the fixed eligible trials, all-pair spectral features, source metadata, raw audit ranges, and extraction code for ds004473 version 1.0.2. They enabled the new chronological test of threshold mixtures under assumed command rates and costs.

• usesShared line-frequency estimation increased mean Brier error by 0.181 in 500-sample buffers from eight SEEG records with different pair frequencies and added-signal RMS 0.25× training RMS

  This source established the retained trial sample, retrospective signal-quality exclusions, and eight training-selected adjacent bipolar pairs. Those fixed choices were inherited through the direct parent and remained unchanged in this study.

• usesStored band covariance kept fixed state decisions equal across two invertible reference changes on matched contacts in eight offline SEEG tests

  Its original pair choices, task labels, trial selection records, and native power features defined the inherited cohort. We preserved its selection logs and checked native features against its stored values.

• usesAn eight-pair SEEG test did not establish lower response detection accuracy on 33 cue–key conflicts than on matched trials in six patients

  This work supplied the original candidate bipolar power features, signal checks, source byte-range records, and extraction method. We recovered them through the direct parent archive and preserved its extraction and metadata retrieval scripts unchanged. Fresh source ranges reproduced all tested conflict and control features.

• usesUncertainty sampling had higher mean later key probability error than random sampling at 64 labels with fixed linear readers in eight offline SEEG records

  This study uses the all-pair SEEG feature cache, source trial records, pinned OpenNeuro metadata, EDF range hashes and extraction code. The package keeps the exact pre-window values and tests a new label-query rule.

• usesKnown simulated contact loss training reduced mean worst-contact Brier error in eight offline SEEG tests with unchanged surviving inputs

  We reused its band power features and pinned raw range retrieval code through the immediate input package. We preserved the recipe and checked saved and fresh voltage samples.

• usesOffline SEEG label sets had 20.2 percent mean single-label error at a nominal 90 percent coverage target in eight recordings

  This publication supplied the exact SEEG power features, trial exclusions, source metadata, and pinned raw retrieval records. We verified the imported hashes and fitted all new decoders from their training trials. We preserve the source code and the raw retrieval recipe.

• usesA 5% rest calibration rule gave 3.9% mean later rest activation and 26.1% task window detection in eight SEEG recordings

  We used its eight-participant SEEG features, source records, retrieval code, raw checks, and model family. We added separate rest calibration and a later operating-point test.

• usesPower blocks with 50 percent overlap increased correct SEEG outputs with a guarantee for a fault of ten samples on one contact in eight recordings at C = 1

  This study holds the inherited early pair ranking, event state definitions, and trial exclusion rules fixed. The archive preserves the original selection code for inspection. It does not repeat the selection.

• usesAdding inter-trial SEEG power increased previous-key loss in eight fixed linear later-trial tests

  We used the complete candidate power arrays, source metadata, and pinned retrieval code. We added a previous-key target, source-neighbor joins, context comparisons, raw checks, and a field audit.

• usesA 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 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.

• usesFive of seven local SEEG votes kept the clean majority under one shaft fault in eight recordings

  We reused its fixed SEEG features, retained trials, splits, source metadata, pinned byte-range records, retrieval code, and model states. These files made the new seven-shaft vote test and source checks reproducible.

• usesJoint Gaussian score integration increased mean command log loss by 0.00003 nats per window after one known simulated contact loss in eight offline SEEG readers

  This publication first preserved the feature arrays, extraction code, trial exclusions, and pinned raw source records used in this study.

• usesCorrect trial matching did not show lower later probability error for conditional logistic readers in eight SEEG recordings

  We reused and verified all eight feature arrays, trial records, source metadata, raw range records, and the source extraction code. We tested a new conditional logistic fit and random trial matching control with separate early calibration.

• usesLater response windows gave lower mean Brier error at both fixed rest positions in eight SEEG recordings with three DPSS tapers

  This work defined the inherited trial eligibility, offline voltage screen, early-trial pair ranks, selected eight bipolar pairs, task-state labels, and chronological train and test split. Those choices stayed fixed in the new window-position comparison.

• usesAdding signed power changes within windows raised mean later log loss at C = 1 in eight SEEG recordings

  This study keeps the eight selected bipolar pairs, early and later trial split, reference power values, and original model record. The later trials are reused. The new test adds signed half-power differences to fixed logistic models.

• usesTraining with simulated four-sample contact delays reduced later worst-contact Brier error in eight SEEG state readers

  This work keeps the retained trial set, native candidate-pair features, chronological split, and training-only pair ranking from the benchmark. It compares new readers within that fixed scope.

• usesRanges from each buffer reduced eight-bit SEEG probability changes compared with fixed early ranges in eight recordings

  We retained its early-trial bipolar pair choices and trial exclusions through the immediate source package. We made no new pair or trial selection.

• usesAdding an intercept to scale calibration reduced later log loss in five of eight SEEG recordings at an assumed 1% task rate

  This publication supplies the complete retained trial features, event metadata, EDF headers, raw audit samples, and pinned extraction code used here. All 129 imported files match its archived hashes.

• usesCalibration trial reuse for reader fitting reduced mean trial coverage by 5.5 percentage points on later trials in eight SEEG recordings at a 90% target

  Used the exact candidate-pair feature arrays, eligibility rules, source sidecars, pinned retrieval recipes, and feature extraction code. All 118 inherited files matched the archived hashes.

• usesA rest reference based on the mean of three power bands gave 3.21 percentage points higher mean later SEEG balanced accuracy than full subtraction of one rest window in eight recordings

  Its original feature recipe defined the power bands, task and rest windows, bipolar pairs, and source quality rule inherited through the rest reference package.

• Adding alpha and beta power to high-gamma power at the same eight pairs did not show a clear mean error reduction in later trials from eight SEEG recordings

  [Legacy relation: reproduces] Fixed C=1 all-band fits and separately selected high-gamma fits repeat the source balanced accuracy and AUC within 1.12e-16. We do not repeat its other controls.

• usesAdding alpha and beta power to high-gamma power at the same eight pairs did not show a clear mean error reduction in later trials from eight SEEG recordings

  This publication supplied all 2014 trial features, exclusions, pinned raw audit ranges, source metadata, and the later-trial split. We use them to compare power-band inputs at eight pairs.

• usesRandom window splits gave 50.9% mean balanced accuracy for random trial labels in eight SEEG recordings with no shared samples

  This study uses source metadata, fixed trial rules, and EDF feature code through the direct parent. They permit an exact source range check.

• usesRest calibration reduces mean task accuracy in fixed SEEG imagery tests from two people

  We adapted its training-only signal input ranking and small logistic reader design. This enabled a fixed reader with eight selected amplifier inputs for the new actual-to-imagined grasp test. We used new source data, four power bands and separate test blocks.

• usesA 32-trial rest covariance update did not establish higher SEEG balanced accuracy than a mean update in eight people

  Used its ds004473 version 1.0.2 log-power arrays, event definitions, eligibility rules, pinned source ranges, and extraction recipe.

• usesShared contact cells certified 58.6 percent of later SEEG decisions for constant contact gains within five percent in eight recordings

  Uses the original rest versus pre-response fixed-reader design, eight training-selected adjacent pairs, candidate feature rules, and later-window split through the direct parent.

• usesA direction rule changed mean key Brier error by -0.00036 for eight SEEG pairs after 64 calibration trials in eight people

  Its feature extraction code and public SEEG provenance support the inherited input features and the independent source voltage checks.

• usesLoss of packets with large voltage changes did not show a clear mean error rise over mask reassignment with linear repair and intact training in eight reused SEEG records at 20 percent loss

  Original trial and contact selection and EDF reader lineage inherited through the direct input package. These fixed pair lists, event anchors, and EDF parsing rules enable the fresh public-source range audit and complete retrieval recipe.

• usesWorst-mask SEEG training reduced mean worst-shaft log loss at C = 1 in eight tests with 32 fit trials

  The reused all-pair features and inherited extraction code come from this feature build. Its source provenance supports raw signal reconstruction.

• usesCorrection with unlabeled rate estimates gave higher Brier error than known rate correction in eight SEEG tests at a simulated 10 percent task rate

  We use its public SEEG features, trial records, source metadata, and range retrieval code. We fit new models on four separate trial blocks.

• usesThirty-two spread calibration trials gave lower mean later balanced accuracy than all early trials in eight SEEG recordings

  This test used the exact eight-patient feature arrays, trial records, source metadata, raw audit fragments, and pinned retrieval code. A separate C=1 fit reproduced all eight reference accuracy scores. Each new calibration set selected and fitted its own model.

• usesAgreed task labels from separate SEEG shaft groups had 31.0% mean error after condition changes in eight recordings

  This source study defined the bipolar power features and inherited trial checks reused here.

• usesEight SEEG key readers gave 53.4 percent mean balanced accuracy after condition changes with 32 calibration trials

  We used the original candidate power arrays, pinned OpenNeuro metadata and range hashes, and source EDF extractor carried through the selection audit. They supplied three power bands and the complete contact-pair candidates. A new full retrieval checked the arrays and all source range hashes.

• usesLoss training increased mean balanced accuracy by 0.34 points under simulated loss of one contact with mean replacement in eight SEEG decoders

  We use the preserved SEEG features, source records, and extraction code. We reproduce the standard eight-pair model before the new contact loss and loss training tests.

• usesAdding band correlations increased mean later log loss for rest and pre-response states in eight SEEG recordings

  This study supplied the original early power choices, retained trials, and EDF range helpers. We kept its eight pairs per person and all retained trials. These fixed choices define the tested scope.

• usesPair selection with test labels raised random label accuracy to 60.3 percent in eight SEEG recordings

  We copied all eight ds004473 version 1.0.2 feature files, trial records, source metadata, and extraction code. These inputs enabled the matched-size pair selection audit. The parent used earlier-trial selection; our invalid test-label rule is a new audit condition.

• usesEight SEEG tests did not establish 50 percent precision under binomial bounds that assume independent trials, stable class distributions, and a 1 percent task rate

  We used all retained feature arrays, trial records, source metadata, raw range recipes, and extraction code from this publication. We fitted new models in separate time blocks and added threshold calibration and exact precision bounds.

More citing publications

First page of citations