A paired test did not establish higher minimum SEEG accuracy from eight pairs without shared contacts under known simulated loss of one contact in eight sessions
We tested whether eight bipolar SEEG pairs without shared contacts improve minimum accuracy during known simulated loss of one contact. We used OpenNeuro ds004473 version 1.0.2 and eight sessions. Selection and fitting used early trials. Later trials supplied the test. Both methods used 24 features. The new rule used 16 physical contacts; the baseline used 11–15. Mean minimum balanced accuracy rose from 65.51% to 66.40%. The paired change was +0.88 percentage points. Its 95% participant bootstrap interval was −1.08 to +2.74 points. The exact two-sided p value was 0.40625. The test did not establish an accuracy gain. The rule limited each contact loss to one selected pair. All eight raw feature rebuilds exactly matched the saved arrays. The package preserves code, full candidate features, source hashes, fitted models, all case predictions, three figures, and reproduction instructions. This offline test does not establish online BCI performance or brain stimulation effects.
stereo-EEG · SEEG · brain-computer interface · contact loss · bipolar signals · feature selection · balanced accuracy · logistic regression · fault tolerance · offline evaluation
References
5- usesLoss training raised mean accuracy by 0.34 points under simulated contact loss in eight SEEG decoders sci:sha256:884a92921286ae95392544128af6a4f343937c3b760e03bce164eec4599853d4
We retrieved its preserved source package and used its physical contact-loss rule with training-mean replacement. We tested selection without shared contacts and used a common union of possible failed contacts for both methods.
- Eight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
[Legacy relation: reproduces] We recovered all eight baseline model selections, scalers, coefficients, and intercepts to 1e-12. The resulting baseline mean intact balanced accuracy was 70.6865%. This reused-data check is not an independent data repeat.
- 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 used the full candidate pair features, windows, trial exclusions, source metadata, extraction code, and chronological split. We tested a new contact selection rule on the same data.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
This study used OpenNeuro ds004473 version 1.0.2; the preserved feature arrays were rebuilt from pinned raw ranges.
- doi:10.1088/1741-2552%2Facae0a
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesA 32-trial recent update did not establish higher final test accuracy than an early update with eight SEEG pairs in eight recorded sessions
We imported its full candidate features, trial metadata, source records, and extraction code. All 118 copied input files matched its archive. Fresh raw audit ranges matched in all eight patients. A new complete feature rebuild for sub-11 matched every array. We test a new equal-budget labeled update with fixed pairs.