A correction from eight known rest windows kept 69.7% mean accuracy under simulated constant bipolar gains of 0.5 to 2 in eight SEEG recordings
Eight known rest windows set one score correction for each fixed SEEG model. We used public OpenNeuro ds004473 version 1.0.2 from eight patients. Early trials trained the models. Separate later rest windows set the correction. The final test used 610 other trials to separate rest from a state before a response. Each simulated voltage gain acted on a bipolar output and stayed constant through calibration and testing. The allowed gain was 0.5 to 2 at each pair and was the same at all frequencies. The correction kept 69.7% mean balanced accuracy, with a descriptive 95% patient interval of 62.4 to 77.4%. Without correction, the exact mean minimum was 49.7%. The gain over power ratios was 2.8 percentage points, with an interval of -1.2 to 7.4 points. This test did not establish better accuracy than ratios. Eight rest windows contained about four seconds of signal but spanned 41 to 69 seconds in the task. The result concerns fixed linear models and simulated constant gains in these offline recordings. The package preserves code, features, source records, decisions, gain proofs, figures, and a successful 84-file numerical repeat check.
SEEG · brain-computer interface · gain calibration · rest calibration · power features · balanced accuracy · signal decoding
References
4- usesBand power ratios retain 67.0% balanced accuracy under simulated constant voltage gains per shaft in eight SEEG recordings sci:sha256:fa4b2be456049aafc97572ea0a4d44239abb9b5dc00341d8bcb7a0fd368bf201
This gain study supplied the related ratio result and the proposed scalar calibration correction. The current study tested that correction with eight separate known rest windows, a final held-out trial set, and a calibration count curve. It does not provide evidence from an independent patient sample.
- 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 archive supplied the retained trial features, source retrieval records, extraction code, and absolute-power baseline models. The current study reused its eight-patient sample and checked the baseline before testing a separate rest correction.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
Source dataset OpenNeuro ds004473, version 1.0.2, used for the eight SEEG recordings.
- doi:10.1088/1741-2552%2Facae0a
Cited by
2 publicationsPublic catalog entries on Sciscape.
- 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 work established the known-rest mean score correction and its constant gain identity. We extend the location rule to the median and test two recorded task replacements and an arbitrary two-score bound. Its different model splits prevent a matched accuracy comparison.
- usesAn eight-trial rest mean raised later SEEG state accuracy by 5.05 points over one rest window in eight recordings
Its correction from eight known rest windows set the gain-stability question. We tested a reference that updates every eligible trial, proved fixed-gain cancellation, and measured residuals from defined gain changes.