Linear filling gave 74.2 percent balanced accuracy for random labels in eight SEEG rest recordings with 20 percent sample gaps set by label
We tested whether gap filling can give false evidence of SEEG label decoding. We used rest voltage from eight people in OpenNeuro ds004473 version 1.0.2. Each person had 100 random label draws. The labels were independent of the original voltage. Ten simulated gaps removed 100 of 500 samples from label 1 windows. Linear filling then gave 74.23 percent mean balanced accuracy on later windows. Gaps independent of labels gave 50.07 percent. The paired difference was 24.16 percentage points. Its descriptive 95 percent person bootstrap interval was 16.22 to 33.46 points. All eight people had a positive difference. Reversing the gap rule at test gave 26.48 percent accuracy. Intact windows gave 50.03 percent. The classifier used band power and received no gap flag. Hold and mean filling gave larger coupled effects. These are offline tests with simulated gaps. They show a measurement path for random label information. They do not show a neural stimulation effect or better clinical control. The package includes exact source windows, source hashes, labels, masks, models, scores, three figures, code, and repeat instructions. A fresh audit matched 16 pinned raw byte ranges. Independent numerical checks and a full repeat passed.
stereoelectroencephalography · gap filling · balanced accuracy · random labels · band power · brain-computer interface · measurement bias
References
7- usesA causal filter gave higher fixed decoder accuracy than fourfold sample removal alone in eight SEEG recordings sci:sha256:e48923d80e223e23c65d73b7a5e6734993070247d942334dabdcfd1ffa26c7b6
The direct parent supplied exact bipolar voltage windows with history, trial and pair records, frozen task models, native features and probabilities, pinned source metadata and range hashes, and MIT retrieval code. These inputs support the new random-label gap test. Preserved files matched its archive hashes.
- usesZero-centered 8-bit waveform codes reduced frozen SEEG task-state accuracy from 70.7% to 50.6% in eight recordings sci:sha256:f11a99582db0d0fff28986ef433b9cde1661cb1668cb5772f0a78262582c2b8c
This work recovered the upstream exact bipolar waveforms and source range records. We used them through the direct parent and made a fresh audit of two source ranges per person.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
This work made the original eight-pair selection, early and later trial split, retained trial rules, band features, and logistic task model. We used those inputs through the direct parent. The secondary native task scores were reproduced.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analyzed dataset is the recorded OpenNeuro ds004473 version 1.0.2.
- doi:10.1088/1741-2552%2Facae0a
- doi:10.3389/fncir.2012.00087
- doi:10.3389/fncir.2013.00006
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesEqual 20 percent sample loss with linear filling gave 56.7 percent balanced accuracy for random labels in eight SEEG recordings
This parent supplies exact selected source windows, frozen task models, source metadata, and retrieval code. We extend its random label gap test to two patterns with equal sample loss and a fixed common mask.
- supportsA fixed 20 percent sample mask reduced mean random label balanced accuracy from 73.5 to 50.0 percent in eight SEEG rest recordings with simulated gaps
A new random label seed repeated above-chance scores when labels set short gaps and below-chance scores when the test rule was reversed. The fixed-mask correction and equal-count layout controls extend the measurement-path evidence on the same cohort.
- usesA fixed 20 percent sample mask reduced mean random label balanced accuracy from 73.5 to 50.0 percent in eight SEEG rest recordings with simulated gaps
The direct parent supplied exact integer SEEG voltages, pair and trial records, source metadata, fixed task models, range receipts, and numerical helpers. Its random label gap test motivated the new fixed-support correction.