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

1. 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 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.

2. supersedesEight fixed SEEG decoders scored 65.8% mean balanced accuracy for response versus rest windows after task condition changes sci:sha256:d06cb877ecdb4aef41ec99088819e4bd46c7ae15cbe48fb22e8419ea7aacc405

   This package corrects the file manifest check after sci fetch adds service metadata. Data, model parameters, predictions, scores, and figures are unchanged. The corrected checker excludes .autosci-publication. The original checker remains in provenance.

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

   The analysis used one recording per person from OpenNeuro ds004473 version 1.0.2; the saved version-specific dataset metadata identifies this DOI and CC0 license.

4. supportsdoi:10.1088/1741-2552%2Facae0a

   The cited source paper describes the forced two-choice task and clinical contact placement for the recordings.

Public catalog entries on Sciscape.

• 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

  We reused its complete SEEG feature arrays, trial records, source metadata, raw audit ranges, retrieval code, and file hashes. We fitted new readers and compared band gains before and after responses.

• usesA score interval gate kept accepted labels fixed under simulated constant voltage loss of up to 6 dB on one shaft in eight SEEG readers

  This archive is the origin of the saved primary readers, feature extraction, and split. We inherited these values through the direct confidence-gate source archive.

• usesFixed signal distance gave no clear Brier score gain over confidence in eight offline SEEG condition changes

  Its fixed response and rest readers supplied the predictions whose correctness we estimated. The direct source package preserved these readers and their fitting recipe. We independently checked their contact selection and fits.

• usesA 25% random trial audit rate and a delay of five trials increased mean error exposure by 23.9 points against complete immediate feedback in eight offline SEEG tests

  The gate source package preserves the feature arrays, fitted reader values, source extraction code, and source range receipts from this reader study. We use those fixed inputs without a new fit to test feedback schedules.

• 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

  This earlier task-condition transfer test guided the interpretation of early calibration and later test conditions in the same cohort. We used its task-state scope. We did not reuse its fitted models or test split.

• usesA ±6 dB shaft gain bound retained 9.0% of windows after simulated loss in eight SEEG readers with fixed equal contact gains

  The eight fixed readers use its feature selection, standardization, fitting method, and saved coefficients. The package includes verified copies of its source code.

• usesEarly penalty selection reduced 300 ms cue SEEG forecast error compared with penalty 1 in eight reused patient tests

  This study uses its unchanged EDF reader, event join, bipolar pair, range request, and power helpers through the parent package. It preserves the helper MIT license.

• usesAn eight-score median reference kept a mean of 47.4% of later gate outputs fixed under any two reference score changes in eight SEEG readers

  This upstream work supplied the fixed early-condition reader construction and source feature extractor used by the direct parent. We preserve its model values and extractor without fitting or selecting a reader again.

• usesFixed early confidence thresholds reduced mean task state error to 24.5% at 49.6% retention after condition changes in eight offline SEEG tests

  We used this archive for all eight feature arrays, source metadata, primary fixed readers, and split records. It enabled a separate early threshold calibration and a later condition test.

• usesAdding 70–110 Hz power gave an uncertain 0.9-point mean accuracy gain after task condition changes in eight records with eight fixed SEEG pairs

  We reused all eight complete feature sets, source metadata, extraction code, raw audit ranges, and the fixed condition split. A secondary reader exactly reproduced its pair choices and coefficients.

• usesRadial basis SEEG readers did not improve mean accuracy after condition changes in eight recordings with 32 calibration trials

  We used its definition of consecutive retained condition segments for later testing. We used 32 calibration trials and compared two SVC kernels. We did not use its fitted readers or outcome scores.

• usesA fixed ridge model with 300 ms of SEEG after the cue increased mean absolute log response-time error in eight patient tests

  We used the source metadata manifest, fixed EDF URLs, and unchanged EDF reader, range, pair, join, and power helper code. New voltage ranges supplied all cue and earlier-window features.

• usesA confidence margin blocked new output errors and withheld most correct outputs in eight SEEG records under an ideal ±3 dB shaft gain model

  This publication supplies the upstream feature extraction and reader fitting through the direct source package.

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

  This study supplied the early fitting and next-condition test split design. We fitted new readers on separate shaft groups.

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

  We used the time-ordered condition test and the response-versus-rest comparison design. We applied new 32-trial calibration and actual response-key targets. No earlier fitted parameters or test predictions entered the new readers.

• usesA simulated 6 dB shaft voltage loss raised mean accepted label error by 10.5 points in eight SEEG readers with fixed early confidence gates

  Its feature extractor and primary fitted model values were used through the direct confidence-gate parent. We preserve the extractor and pinned source recipe.

• usesEarly fits did not establish lower loss than full rest correction with a 10-trial half-life in eight SEEG readers

  This publication supplies all eight feature sets, primary fixed reader states, trial-fit indices, and source metadata. We reuse these files with hash checks. We add an early bounded correction-strength fit and test the next retained condition.

• usesAdding beta and high-frequency log power products did not show a clear Brier gain after a task change in eight people with eight SEEG pairs each

  We used its source-order task-segment protocol to separate early calibration, later first-condition trials, and next-condition trials. We added nested penalty choice and product controls.

• usesMaps from 16 trials in a new task condition gave 0.028 bits more mean loss than matched earlier maps in eight fixed SEEG decoders

  We reused all eight complete SEEG feature sets, fixed primary decoder states, source metadata, retrieval code, and raw audit ranges. The new test compares confidence maps from matched earlier and new condition labels on a shared later test.