Gaps of 50 samples gave higher later Brier error than gaps of 1 sample with linear repair and training on intact data in eight reused SEEG records at 20 percent loss
Gap layout changed later task-state probability error in eight reused public SEEG records from OpenNeuro ds004473 version 1.0.2. Each periodic layout removed 100 of 500 samples. We tested gaps of 1, 10, and 50 samples at every integer mask origin. With linear repair and training on intact data, mean Brier error rose from 0.185667 to 0.204063 for gaps of 50 rather than 1 sample. All eight records had higher error. The paired increase was 0.018396, with a descriptive 95 percent person bootstrap interval of [0.010790, 0.026123]. Training on the same gaps reduced the 50-sample mean error to 0.186043. Mean repair gave a different order of errors across layouts. Missing sample fraction alone did not specify reader error in this benchmark. The package preserves exact selected signals, code, source range hashes, models, all later scores, three figures, and repeat checks. The test used simulated missing samples and a reused cohort. It did not test stimulation or a live BCI.
stereoelectroencephalography · signal gaps · Brier error · brain-computer interfaces · missing data · signal repair
References
5- usesThree DPSS tapers gave lower mean later Brier error than a Hann taper in eight reused SEEG records with fixed bipolar pairs and 500-sample buffers sci:sha256:bd2b70f0665b6319bf4e85827eb9f5e099defe2ec47485c3b3f543374625d480
This publication supplied all exact selected SEEG buffers, metadata, selection files, retrieval and audit code, and the fixed-C Hann reference. The new comparison keeps those inputs and adds equal-loss periodic layouts, all mask origins, two repairs, and matched training with fixed total fitting weight.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
The new test inherits the original retained trial list, state labels, and eight-pair selection through the direct data source. It makes no new selection.
- usesTraining with gaps increased mean later state accuracy in eight SEEG recordings with 20 percent simulated sample loss sci:sha256:3cd2d827953f8ece1cacf7ff63ec2290c89a8689f1a8694c080435367b8979bc
This earlier gap study supplied the comparison question and the reason to report probability error and score ranking. The new test uses every mask origin, equal-loss layouts, and a different matched fitting recipe.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The study reuses selected SEEG signals and metadata from OpenNeuro ds004473 version 1.0.2.
- supportsdoi:10.3389/fnhum.2022.1016379
This experimental study supplies the background evidence that device blanking settings can change recorded signal distortion, motivating tests of missing samples.
Cited by
6 publicationsPublic catalog entries on Sciscape.
- usesPair repair reduced mean Brier error under equal 60 Hz voltage at one training RMS in eight reused SEEG records with simulated contact loss
This publication supplied the prior gap layout test and reuse context. It simulated gaps after bipolar subtraction. The new study tested contact loss before subtraction and equal added reference voltage.
- usesSine fits did not establish lower Brier error than linear repair in eight SEEG records with matched training and 20 percent periodic loss
This study supplies the earlier gap-training result and 20 percent periodic-loss comparison. We add a separate observed-sample sine-fit map and use different fixed gap positions for training and testing.
- usesSine fits did not establish lower Brier error than linear repair in eight SEEG records with matched training and 20 percent periodic loss
This study supplies the earlier gap-training result and 20 percent periodic-loss comparison. We add a separate observed-sample sine-fit map and use different fixed gap positions for training and testing.
- usesMedian 70–110 Hz power gave higher mean Brier error than block mean power in eight SEEG records with one simulated three-sample contact pulse at ten times early RMS
This study uses the exact digital bipolar buffers, trial and contact selection, source metadata, EDF range reader, and logistic fitting equations from this publication. New code computes gamma-only readers and contact pulse tests.
- 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
Supplied all eight compact SEEG signal files, fixed trial and pair selection, source metadata, MIT power and logistic code, range retrieval code, and the intact-trained Hann reference. All 114 copied input files matched archived hashes. The new analysis adds voltage-dependent packet masks and mask reassignment controls.
- usesPair repair reduced mean Brier error under equal 60 Hz voltage at one training RMS in eight reused SEEG records with simulated contact loss
This publication supplied the prior gap layout test and reuse context. It simulated gaps after bipolar subtraction. The new study tested contact loss before subtraction and equal added reference voltage.