A fixed covariance rule gave higher SEEG site accuracy than a diagonal rule in offline tests from five patient sessions

1. usesA joint SEEG shape decoder did not show a clear mean gain over one selected pair in four patient sessions sci:sha256:81f77d4f52365d5b14fed7d3d6db7b430a2cf478b9a6a6062540c4e90bbd8e7c

   We used this report’s verified voltage bins, source metadata, retrieval and extraction code, pulse eligibility, chronological per-site allocation, and equal-pair shape comparator. We rebuilt its inputs from raw files. Our covariance test uses the same patients and sessions, so it is not independent replication.

2. usesSEEG response shapes at 15–100 ms gave higher site identification accuracy than shapes at 100–500 ms within each of five patient sessions sci:sha256:aa5b71d76d214f861e99964287f9da01f3f9fc6c707a4dcfe24807283918ab40

   We used this earlier site benchmark to choose the pulse-site identification question and response-window controls. It uses the same source patients. We did not treat its evidence as an independent sample.

3. usesdoi:10.18112/openneuro.ds004457.v1.0.2

   The analysis used OpenNeuro ds004457 version 1.0.2; the source data and metadata record CC0.

4. supportsdoi:10.1523/JNEUROSCI.1325-22.2023

   The source study supplies the published context that temporal and limbic stimulation produces distinct responses in ventral temporal cortex.

Public catalog entries on Sciscape.

• usesEnergy shared across four pulses gave no clear offline site-rank gain over total energy in five SEEG sessions at 100 to 500 ms

  This work supplied the inherited raw retrieval and native-sample voltage extraction recipe. We used that recipe for all 110 pinned files and reproduced all five binned input arrays.

• usesCorrection with four pulses kept all 15 site choices in five offline SEEG sessions from 20 to 500 ms

  This earlier work supplied the contact and eligibility rules inherited through the voltage-bin source package. We adapted those rules in the fixed cohort loader. It is an upstream code and cohort source, not an independent validation cohort.

• usesOne shared temporal noise matrix gave lower mean SEEG site accuracy than full covariance with two fit pulses per site in offline tests of five sessions

  This publication supplied the fixed five-session cohort, voltage inputs, and full covariance rule. We used its inputs as a reference for a fresh raw replay. Its four-pulse comparator counts matched our counts in all 30 checks. The new test compares a shared temporal matrix with fixed methods using two fit pulses per site. These are reused patients and pulses.

• usesSEEG on other shafts gave 78.7% mean accuracy for two known neighboring stimulation sites in five offline sessions

  This is the compact-bin source for the inherited extraction and fixed covariance reader. The present work retains its source credit and changes readout selection and shaft exclusion.

• usesA fixed average of local and earlier neighbor pulses increased 20–100 ms SEEG forecast error in three of five sessions

  We used its voltage bins, fixed recording contacts, event rules, and extraction code. A fresh raw rebuild reproduced these inputs. We added chronological donor selection and local-versus-neighbor response forecasts.

• usesBipolar bypass raised mean site accuracy from 28.4% to 31.9% at 20–100 ms after known simulated contact loss in five SEEG sessions with unchanged surviving voltages

  Used the calibration-only site means and fixed half residual-covariance, half diagonal fit. Transferred the frozen model to observable pair combinations after a known contact loss.

• usesSEEG at 20–100 ms gave higher remote site accuracy than shifted controls in four of five offline 6 mA sessions

  We used its fixed half-covariance distance rule and variance floor. We kept this rule fixed at every shift and fitted each state from fitting pulses only.

• usesNine shifted templates increased mean site accuracy under nine equally weighted simulated timing offsets from −20 to +20 ms in five offline SEEG sessions

  We adapted its MIT raw MEF extraction code, supplied through the direct parent package. Fresh raw reconstruction checked all retained contact-bin arrays against the archived inputs.

• usesTwo voltage checks accepted fewer correct responses than one check in five offline SEEG sessions at a 5% background calibration target

  The inherited extraction, retrieval, contact eligibility code, and fixed half-covariance model came from this work through the direct input study. We preserve the code credit and source record. We refit the common model with three fit pulses per site and add background means and response scores.

• usesA 90% calibration voltage map left positive test contrast outside its space in five offline SEEG sessions

  This publication supplied the voltage arrays, source manifest, fixed recording pairs, site selection, and calibration covariance rule. We rebuilt its inputs from the public raw files.

• usesNeighboring SEEG site responses had estimated shape separation of 0.16 at 20–100 ms in five sessions under stable means and independent group errors

  The study reused the exact hash-checked voltage bins, event tables, source metadata, source hashes, and retrieval and extraction code from this package. An independent inclusion implementation matched its physical pulse allocation.

• usesSite variation scales give mixed probability gains in five SEEG sessions with three template pulses and separate calibration

  We used its archived ds004457 version 1.0.2 voltage bins, metadata, target contacts, eligibility rules, raw-file hashes, and extraction code. We adapted its loader and rebuilt its inputs from public raw files. This enabled a paired test of shared and site-specific variation scales.

• usesFour local pulses gave lower 20–100 ms SEEG forecast error than one pulse with disjoint-site projection in all five tested patients

  The fixed recording targets, adjacent contact convention, raw bin extraction, and read_subject selection rules originated here. They were imported through b37f98c3. The new forecast preserves those input rules and adds strict earlier disjoint donor selection.

• usesFull covariance gave no mean gain in offline four-site accuracy after stimulation contact order reversal with four SEEG pairs and five fit pulses per site in eight patients

  Used its fixed 0.5 covariance fraction as the starting method. Changed the diagonal target to the inherited blended variances. Compared the full rule with a diagonal rule that has identical variances in eight STReEF patients. This sample differs from the five ds004457 sessions in the prior study.

• Coordinate spacing gave lower mean eight-site accuracy than random sets in five offline SEEG tests at equal calibration pulse counts

  [Legacy relation: reproduces] Reproduces the full-catalog signed 20 to 100 ms reader counts, 56, 50, 82, 45, and 54. It also supplies the inherited contact choices, input lineage, covariance rule, and source code.

• usesMean voltage from four pulses at the same site gave 43.8% accuracy and label votes gave 40.1% at 20 to 100 ms in five offline SEEG sessions

  We used its fixed C = 0.5 S + 0.5 D distance rule and variance floor, fitted only on early pulses. We tested aggregation without choosing a new covariance weight from test results.

• usesWindows selected from calibration gave higher mean site accuracy than 500 ms windows in five offline SEEG sessions

  We used the fixed cohort, voltage bins, contact list, eligibility code, source recipe, and covariance decoder. This study adds a window choice from calibration data and tests it on separate later pulses.

• usesSeparate pulse groups showed larger 20–100 ms SEEG site distances than pre-pulse distances in five sessions under a fixed noise scale

  Uses archived voltage bins, source metadata, contact selection, pulse inclusion, and raw extraction code.

• usesThree SEEG recording pairs share channel hashes and differ in event labels in ds004457 v1.0.2 and ds004696 v1.0.1

  The summary of this within-session site-decoding test motivated the BCI use of ds004457 pulse-response data. We did not copy its code or model outputs. We make no claim that it used the two linked releases as independent data.

• usesSimulated joint contact gain errors within five percent changed 172 SEEG decisions that passed separate contact tests in five offline sessions

  This archive supplied the exact contact bins, source metadata, readout targets, pulse rules, fixed covariance reader, and raw retrieval records. We checked all ten input hashes and reproduced all 2,196 nominal predictions across three windows. The new result compares separate and joint contact gain tests.

• usesA fixed uncertainty rule gave no clear gain over random repeat requests for known neighboring sites in five offline 6 mA SEEG sessions

  The study uses the inherited fixed covariance reader and source extraction method. A fresh source rebuild checks the compact voltage inputs.

• usesCalibration across five reading times kept more true sites but gave fewer labels in five offline 6 mA SEEG sessions

  Supplied compact voltage bins, source metadata, target contacts, and MIT retrieval and extraction code. This run verified the inherited hashes and rebuilt all inputs from the public raw files. The new code tests calibration across five times with separate fit, calibration, and test pulses.

• usesWith four pulses per site, late calibration gave higher late SEEG scores in five offline sessions

  We used its inherited native-sample extraction and fixed raw source manifest through the immediate input package. Fresh raw downloads and extraction reproduced all ten compact input files byte for byte.

• usesA joint score for 20 to 100 ms SEEG responses reduced mean before-pulse acceptance and increased mean wrong-site acceptance in five offline sessions

  The fixed recording contacts, eligible sites, current rule, and early pulse split come from this earlier study through the direct parent. The contact reader retains these rules.

• usesFrequency weights raised mean minimum site coverage from 63.5 to 89.8 percent in five SEEG sessions when 80 percent of simulated commands used one shaft and site responses stayed fixed

  This is the original compact-bin and raw extraction source inherited through the immediate parent. We retained its contact and event rules and extraction credit. A fresh download of the pinned public subset and raw reconstruction matched every compact data file exactly.

• usesA SEEG reader kept an after-pulse site accuracy gain after removal of linear trend use in five offline patient sessions

  We used the five-session voltage bins, event tables, contact and site rules, four-fit and four-test pulse split, covariance rule, source manifest, and extraction code. We reproduced the ordinary primary inputs, models, and scores. We then tested an exact distance rule that removes linear trend use, with matched before-pulse controls.

• usesNoise correction gave a mean change of +0.39% in forecast error for 20–500 ms SEEG distances in five patients

  This earlier work supplied the fixed contact, site, and pulse selection rules through the immediate source. Its chronology audit informed the within-site interpretation of later pulses.

• usesA five-point fourth-pulse rule lost 7.9 percentage points in one of five SEEG sessions at 20 to 100 ms

  The reused compact data and fixed cohort follow its recording pairs, site exclusions, first-four and last-four pulse roles, and fixed covariance benchmark. We preserved the offline scope and original bin provenance.

• usesOne log RMS value per pair reduced site accuracy in five SEEG sessions with a fixed reader

  We reused its exact voltage files, selected recording contacts, eligibility rules, and pulse split. We used its fixed covariance reader. The signed primary correct counts match the earlier report. The new test compares fixed response representations.

• usesVariation in four early pulses gave lower SEEG error-limit scores than response energy at 20–500 ms in five patients

  The inherited source manifest and raw MEF extraction method originate in this work. We used them to verify and rebuild pinned OpenNeuro ds004457 inputs.

• usesNoise scaling gave 24.5 percent mean site accuracy with four calibration pulses per site on new shafts in five offline SEEG sessions from 20 to 500 ms

  Its compact SEEG inputs and fixed half-covariance rule underlie the parent package. We reused the raw retrieval and extraction recipe, checked all source hashes, and rebuilt every compact input from fresh MEF downloads.

• usesSelection from three windows and one early window gave the same mean lowest site accuracy in five offline SEEG sessions

  We reused its ten compact voltage and event files, source manifest, BIDS metadata, extraction code, eligibility rules, four fit pulses per site, last four test pulses per site, and fixed half covariance reader. We rebuilt all voltage arrays from the public raw source before the new comparison.

• usesAn early uncertainty gate gave no clear gain over random waiting for known neighboring sites in five offline 6 mA SEEG sessions

  This study used the shared covariance score method and the source extraction code. It added a calibration-only early wait rule and an exact random-wait baseline.

• usesFive time templates reduced SEEG site log loss with simulated grid errors up to 10 ms in five patients

  This work supplied the exact source bins, source hashes, retrieval and extraction recipes, contact and site rules, and fixed covariance fraction. We kept these inputs fixed and tested five pulse-time templates on separate fit, scale and later pulses.

• usesNoise scaling gave higher site accuracy at equal ranks in five SEEG sessions from 20 to 500 ms

  We used its ten compact voltage and event files, fixed five-session cohort rules, source extraction, and 50% covariance benchmark. We checked every input hash and rebuilt all inputs from pinned raw files. The new test compares response maps at equal ranks.

• usesFinite template scaling increased pulse use without reducing site error at a 90% threshold in five offline SEEG tests

  The recording and stimulation eligibility rules, pulse grouping, and shared covariance reader are inherited through the source code.

• usesSEEG data after a pulse gave higher site accuracy than data before a pulse for neighboring stimulation pairs in five offline sessions

  This study uses the verified input bins, source contact rules, extraction code, source manifest, and fixed covariance method. It tests neighboring two-site choices that share one contact, rather than all-site choices.

• usesOne waveform learned from two pulses per site missed repeated site differences in five SEEG sessions

  This archive supplies all five patients, hash-checked 5 ms voltage bins, recording pair and site rules, raw extraction code, and source provenance. The new test measures repeat power outside a direction learned from separate pulses.

• usesA 90% timing prior gave no clear accuracy gain over nine shifted SEEG templates in five offline sessions

  We retained its fixed half-full, half-diagonal pooled residual covariance rule. All timing rules used that same covariance construction. The new primary window was 40 to 120 ms.

• usesThree response directions gave lower AUROC than full distance for excluded stimulation shafts in five SEEG sessions

  We retained its full Mahalanobis template distance and fixed 0.5 within-site covariance shrinkage as the baseline. The new test measured distance along three known template directions.

• usesA positive gain fit reduced mean site accuracy from 40.6 to 37.9 percent at 20 to 100 ms in five offline SEEG sessions

  We used its contact bins, event tables, source manifest, selection code, and fixed covariance comparator. All five baseline accuracies matched. We downloaded the public raw files again and reproduced the inputs exactly. The same patients are reused.

• usesAdding a fixed response mean score did not improve average SEEG site accuracy at 20–100 ms in five offline patient tests with four calibration pulses per site

  We used its exact input voltage arrays, source metadata, eligibility rules, pulse allocation, and centered covariance reader as the comparator. All 732 primary centered decisions matched. The same five patients were reused, so this is not an independent cohort.

• usesTwo four-bit storage rules reduced mean site accuracy in five offline SEEG sessions with four calibration pulses per site at 20 to 100 ms

  We used the inherited compact-bin provenance, eligible sites, recording pairs, calibration split, and fixed covariance rule. The immediate source carries these inputs from this work.

• usesA template gap gave 0.827 mean error AUROC across known sites in five offline SEEG sessions at 20 to 100 ms

  We used the fixed pooled within-site covariance and 0.5 diagonal shrinkage rule through the immediate parent. No covariance setting was tuned on this test.

• usesTraining with nine window shifts raised mean site accuracy under uniform −20 to 20 ms shifts but reduced accuracy at zero in five SEEG sessions

  We used its fixed half-diagonal covariance distance rule. Our window is 40 to 120 ms, and our residual moment gives each physical pulse equal total weight across shifted views.

• usesFull covariance did not establish a mean increase in SEEG response spread at 20 to 100 ms in five sessions under stable means and independent errors with zero mean

  Used its fixed covariance comparator C = 0.5 S + 0.5 D and diagonal comparator. Applied both metrics to separate late-pulse response moments rather than treating site accuracy as a spread measure.

• usesA fixed pre-pulse adjustment gave 70.8% accuracy versus 71.4% without adjustment for neighboring sites at 20–100 ms in five offline SEEG sessions

  We used its fixed half-diagonal covariance rule for the response and joint models.

• usesTwo calibration pulses gave a mean score gain of 0.062 bits over eight sites for 20 to 100 ms SEEG responses in five sessions

  We used its fixed half-covariance reader rule through the cited input package. The new analysis retained voltage amplitude and used separate probability fitting, size choice, and final test pulses.

• usesA relative distance score gave higher mean AUROC but accepted most SEEG responses from withheld shafts in five offline sessions

  Uses its fixed covariance shrinkage rule for the conditional and background response models.

• usesTwo response baselines gave opposite mean SEEG associations with pre-pulse voltage change in five patient sessions

  Its original bin extraction code and fixed contact list enabled an exact check against newly retrieved raw recordings.

More citing publications

First page of citations