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.

• usesTest labels in pair selection raised mean random-label SEEG accuracy in eight recordings with 32 calibration trials

  Uses the unchanged eight SEEG feature archives, trial exclusions, source metadata, pinned raw audit ranges, and extraction code. The full-calibration training-only models reproduce all parent test decisions. The parent selected pairs from training trials only; the new experiment deliberately adds a test-informed selection control.

• usesAdded spatial power gave no clear mean test error decrease with pairs selected from band contrasts in eight SEEG records

  We used the preserved log-power features, trial rules, raw audit bytes, source metadata, and range extraction code. They enabled the later-trial reading test and source checks.

• usesA 20-trial delay in correct feedback changed mean absolute coverage error by 0.23 percentage points in eight SEEG replays at a 90% target and a 0.05 step

  Used the candidate-pair feature arrays, eligibility rules, BIDS sidecars, version-pinned byte recipes, and feature extraction code. These inputs support the new feedback replay.

• usesRemoving the test correction did not give a 10 percent error gate in eight offline SEEG recordings under a binomial trial model

  We used its complete public SEEG features, trial records, source metadata, and version-pinned raw byte ranges. We checked 118 inherited file hashes and independently checked stored and fresh source ranges.

• usesShifts after bipolar subtraction reject 99.55 percent of shared contact beta tests at a five percent threshold with independent contact shifts from eight SEEG records

  We used its fixed EDF headers, metadata, source object version URLs, and MIT retrieval and header helpers. We read new byte ranges and added a contact-null calibration analysis. We did not use its decoder features or results.

• Removing the check after the response restored three trials and reduced observed mean later balanced accuracy by 0.043 percentage points in eight fixed SEEG readers

  [Legacy relation: reproduces] Replayed the original later-trial eight-pair reader scores. An independent refit kept every class decision and reproduced mean balanced accuracy at 70.686490%. This is a same-cohort numerical replay, not an independent-cohort replication.

• usesRemoving the check after the response restored three trials and reduced observed mean later balanced accuracy by 0.043 percentage points in eight fixed SEEG readers

  Used retained features, original eight-pair reader states, trial and exclusion records, source metadata, raw-range hashes, and MIT-licensed extraction functions. Recovered all 18 voltage-rejected trials and audited four nested checks with fixed time limits.

• usesGamma envelope concentration lowered mean later-trial Brier error against three band powers in eight reused SEEG recordings with 500-sample buffers

  We retained its early-trial contact selection, trial eligibility, event-aligned states, and early/later split. The new feature comparison is conditional on that fixed design.

• usesThree-bit SEEG power codes used 90.6% fewer feature bits with a 0.59-point mean accuracy loss in eight offline later-trial tests

  We used its byte-identical SEEG features, trial exclusions, source retrieval code, chronological split, and reference models. The new study tests fixed feature codes and their accuracy loss.

• usesRecent sample weights gave lower mean Brier error than reversed weights in eight reused SEEG recordings with 500-sample buffers

  Provides the inherited eight bipolar pairs per person and offline trial eligibility rules. The new study keeps both choices fixed.

• usesBeta selection gave lower high-gamma reader accuracy than gamma selection in later trials from eight SEEG recordings

  This source publication supplied the retained SEEG features, trial records, raw audit samples, and retrieval code. The new study verified 122 copied files against its manifest. These files enabled the disjoint trial map test and the fixed eight-pair gamma reader comparison.

• usesEarly loss tests did not show better contact priorities than coefficient size under simulated loss in eight SEEG recordings

  This study uses its exact SEEG power features, trial eligibility, source metadata, raw audit ranges, and extraction code. It trains new decoders and tests contact priorities in separate time blocks.

• usesSeparate allowed 12-sample errors at the window center changed all 1,348 tested SEEG decisions within a 100 µV RMS limit per pair in eight people

  This work supplied the original early pair choices, trial choices, split, reference features, and EDF helper bodies through the data parent. We kept these choices fixed.

• usesInput selection by effect agreement changed mean later state Brier error by −0.0008 with 64 trials and eight pairs in eight SEEG recordings

  We used its band-power features and preserved extraction code through the immediate source package.

• usesMean removal before synthetic linear gain ramps reduced decision changes in eight fixed SEEG contrast readers

  We reused all 2,014 retained trial features, source metadata, archived byte range receipts, and extraction code. We fetched and checked 674 later-trial raw ranges for each gain condition.

• usesAdding 70–110 Hz power raised mean balanced accuracy from 65.7% to 68.9% for windows before responses versus rest in later trials from eight SEEG records

  This work supplied the original public SEEG feature extraction and trial rules. We used these inputs and the raw retrieval code through the parent publication. Our study adds a fixed comparison of bands and response periods.

• 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 original feature extraction defines the bipolar pairs, five source windows, three power bands, and trial exclusions used here. We recovered these files through the direct parent archive. We use a new split, a fixed 32-trial selection set, and recent or early added labels.

• usesSpectral contrasts gave 61.2% mean balanced accuracy under fixed shaft gains within ±3 dB in eight SEEG recordings

  Its original extraction code, pinned public source receipts, EDF headers, and raw signal excerpts supplied the provenance and independent raw signal checks preserved through the feature publication.

• usesA shared SEEG model gave 52.6% mean balanced accuracy in eight tests that held out one person after rest calibration

  Used all eight preserved SEEG feature arrays, trial metadata, source records, exclusions, and raw retrieval code. Added a shared model that holds out each person’s task data. Reproduced the parent eight-pair balanced accuracies as a reference.

• usesA fixed rule based on taper spread gave a mean Brier change of -0.00072 in eight reused SEEG records with buffers of 500 samples

  This study keeps the trial and bipolar pair rules from this source. The direct input package preserves those rules and the resulting pair lists. No new trial or pair selection is made.

• uses64 matched calibration trials increased mean later SEEG balanced accuracy over 16 with a fixed penalty in eight offline recordings

  This study supplied the verified SEEG features, event windows, later test, retrieval code, and source records. We reused these files.

• usesTwo 32-trial calibrations within one condition shared 2.5 of eight state pairs and 0.625 key pairs on average in eight SEEG recordings

  We use its three-band features, 500-sample windows, adjacent bipolar pairs, extraction code, and pinned source hashes. We keep these choices fixed for the map comparison.

• usesRefitting with saved data gave a small and uncertain gain after simulated loss of one contact in eight SEEG decoders

  This original archive supplied the feature definition, eight-pair selection, chronological split, and standard decoder through the cited contact-loss archive. We kept that pipeline fixed for the new recovery comparison.

• usesIndependent trials and separate patient/key rates limit monotone 95% mean balanced accuracy bounds to 82.7% for 32 SEEG conflicts in five patients

  The conflict benchmark includes this work’s retained SEEG feature arrays and metadata. I kept them for the new label audit, fresh raw checks, and full reproduction recipe.

• usesTargeted feature changes on one shaft caused wrong outputs from all 507 withheld windows in eight fixed SEEG output gates

  This work supplies the feature files, trial rules, time split, and baseline reader. We used these inputs and reproduced the full reader before we tested the shaft omission gate.

• usesEndpoint tests missed seven decision changes between simulated contact gains 0.8 and 1.2 in eight SEEG records

  We use its original OpenNeuro feature extraction, retained trial rules, event windows, source metadata, and pinned raw range records, as preserved in the contact gain package.

• usesA 64-window rolling rank map raised mean Brier error in eight later SEEG tests

  We used all eight feature sets, trial records, source range hashes, and the extraction code. A fresh public raw-range rebuild matched every saved feature value. These data support the new rolling-map and replay comparisons.

• usesA rest threshold set for 5 percent errors gave 3.9 percent false detections and 26.5 percent sensitivity in eight later-trial SEEG tests

  We reused all eight people, retained trials, bipolar spectral features, extraction code, and raw range provenance. We fitted new models with separate training, calibration, and later test periods to measure the false detection and sensitivity tradeoff.

• usesAdded imaginary coherency increased mean Brier error for tuned linear readers in eight SEEG records with fixed 500-sample inputs

  This study supplied the original training-based pair selection, trial eligibility rules, and source event clocks. These choices stayed fixed in the new comparison.

• usesIndependent trials and separate patient/key rates limit monotone 95% mean balanced accuracy bounds to 82.7% for 32 SEEG conflicts in five patients

  The conflict benchmark includes this work’s retained SEEG feature arrays and metadata. I kept them for the new label audit, fresh raw checks, and full reproduction recipe.

• usesUsing 100 instead of 500 samples for high gamma raised mean Brier error on later trials in eight SEEG readers with fixed pairs, Hann power, and 500 samples for lower bands

  This study supplied the inherited early-trial pair ranking, eight pair choices per person, trial quality rule, and later-trial split. Its code and publication record are preserved for inspection.

• usesSame-trial power subtraction did not show a clear Brier error gain for eight SEEG pairs on later blocks in six recordings

  We used its complete SEEG power features, source metadata, raw retrieval code, and independent raw check functions. These inputs enabled the new actual-key test and the full six-person public raw rebuild. The new analysis uses different labels and block splits. It does not reuse the source reader fits.

• usesLater trials in eight SEEG recordings have mean 2.4% rest alarms and 9.5% response detection after five-window calibration at a nominal 5% level

  This work supplied exact retained trial identities, original spectral features, source metadata, raw audit samples, and MIT-licensed extraction helpers. We added three rest windows, changed the chronological split, and fitted new models and alarm thresholds.

• usesEight fixed SEEG decoders scored 65.8% mean balanced accuracy for response versus rest windows after task condition changes

  We reused the retained SEEG power features, trial records, exclusions, raw audit ranges, source hashes, and retrieval code. We made new forward condition splits and fit new fixed models. No parent model parameters were used.

• usesContact selection before cross validation raised estimated balanced accuracy by 15.6 percentage points in eight SEEG recordings with 32 calibration trials

  Uses the exact 2014-trial features, source exclusions, extraction code, and signal retrieval records.

• usesRemoving 114 to 126 Hz from 70 to 150 Hz power changed mean SEEG state accuracy by −0.36 percentage points on later trials in eight recordings with fixed pairs and refitted readers

  This study defines the inherited trials, eight early-selected pairs, rest and before-response windows, and native 70 to 110 Hz reference. We keep its selection and later-trial split. The new tests are conditional on those pairs.

• 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.

• usesSelection for the highest surviving score showed no clear Brier error decrease after known shaft loss in eight later-trial SEEG tests at 16 contacts

  Use its eight SEEG feature arrays, fixed source trial sample, extraction code, pinned EDF range receipts, and metadata. All 121 copied input files matched its manifest hashes. Fresh source checks passed in every record; a full sub-11 rebuild matched every array.

• usesConstructed rest score changes gave equal score distributions at 1 and 10 percent task rates for eight offline SEEG readers

  We use the original bipolar SEEG feature method and extraction code. Fresh public signal ranges check 160 feature windows from all eight recordings.

• usesAll available SEEG pairs had an uncertain mean task-state accuracy gain over eight pairs after 32 calibration trials with L2 C=1 in eight records

  We reused the exact public power features, trial records, source metadata, raw audit ranges, and retrieval code. We made new model fits, input-count comparisons, short calibration splits, penalty checks, and figures.

• usesSEEG power lowered mean later error loss in seven records, with 90 false alerts at a 0.05 probability threshold

  Used the original bounded EDF feature extraction method through the cited label audit and block test. Preserved its source extraction code and versioned byte-range recipes.

• usesWith a Hann window, three phase-agreement features did not lower mean later-trial Brier error in eight selected SEEG records

  This study retains the original contact and trial selection. It uses the same rest and before-response labels in eight records.

• usesA Hann gamma envelope feature did not lower mean Brier error beyond matched power on later trials in eight reused SEEG records

  This work supplied the inherited trial eligibility and early-training contact-pair selection. We kept its eight pairs per person and preserved the selection and exclusion records.

• usesThree forward validation blocks did not show a clear Brier error reduction for separate power and phase penalties in eight fixed SEEG records

  Keep the original contact selection, retained trials, exclusions and chronological outer split. The inherited selection records and publication manifest are preserved.

• usesPast rest subtraction gave 58.5% mean balanced accuracy compared with 61.2% for spectral contrasts in an offline state test on eight SEEG recordings with 32 fit trials

  We used the complete feature cache, public data retrieval code, trial records, metadata, and raw range records. These files made the new past rest comparison possible. We preserved their hashes and checked raw examples against the public source.

• usesSigned gamma timing gave an uncertain Brier error change in eight reused SEEG recordings with 500-sample buffers

  This study supplied the original retained trial list, early pair selection, event-defined states, and source extraction code. The new comparison is conditional on its fixed trial and pair selection. The package preserves its code as reference files.

• usesTen spaced gaps gave lower mean zero-fill SEEG accuracy than a circular gap block at 20 percent simulated sample loss in eight fixed decoders

  The original eight-pair decoder supplies trial selection, temporal splits, pair ranking, the three power bands, feature scaling, fixed logistic models, and the native 70.69 percent mean balanced-accuracy reference. Native later-test features and probabilities were reproduced.

• usesPost-response training gave no clear mean phase AUC gain after 32 fit trials in eight offline SEEG readers

  We used all exact SEEG power features, retained trial records, exclusions, source receipts, and extraction code to fit and check the new phase readers.

• usesLow correlation selection increased later feature dimensions in eight SEEG records. Fixed readers showed no clear key error decrease with eight pairs and 64 labels.

  This publication supplied the original five-window feature source and extraction code. We preserved its source metadata, pinned raw range recipes and code. We rebuilt all 202 raw trials for one person and checked sampled trials for all eight people.

• usesPower ratios kept 66.9% mean balanced accuracy under simulated constant ±6 dB bipolar channel gains in eight offline SEEG tests

  We used all 2,014 retained trial features from the eight source recordings. We reused its source records and retrieval code. We reproduced its selected eight pairs, fitted absolute models, and individual later-trial scores. These fixed inputs enabled the same-pair ratio comparison and exact bounded-gain test.

More citing publications