Matched-trial fitting gave no clear gain in later single-window Brier error in eight SEEG recordings after 32 early trials
We tested matched-trial logistic fitting for a later task-state reader after 32 early calibration trials. We used public OpenNeuro ds004473, version 1.0.2. Each reader used eight bipolar SEEG pairs and three power bands. Pair selection, scaling, and penalty choice used early trials only. The later test used 674 trials from eight people. Mean single-window Brier error was 0.21941 with ordinary fitting and 0.21967 with matched-trial fitting. The mean error reduction was -0.00026. Its descriptive 95% person bootstrap interval was [-0.00188, 0.00135]. Four people had lower matched-trial error; four had higher error. The results did not establish a gain or equal methods. A separate pair-choice test used both windows and a known trial structure. It does not show continuous intention reading. We release the input features, source records, code, fitted models, predictions, figures, and independent checks. The complete rerun matched every analysis output. This study does not test brain stimulation.
stereoelectroencephalography · matched-trial fitting · conditional logistic regression · Brier error · neural decoding · model comparison · calibration trials · power features
References
4- usesContact selection before cross validation raised estimated balanced accuracy by 15.6 percentage points in eight SEEG recordings with 32 calibration trials sci:sha256:2a94c909e0019de38093ca2e8167cc4b87ac1b4d7532223ee7271179ec4267f0
This audit informed our rule to select pairs and scale inputs inside each early validation fold. We do not use later test trials to choose the penalty.
- 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 use all eight unchanged SEEG feature files, trial records, source metadata, and pinned raw-range records. These files support the new matched-trial and ordinary fitting comparison.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis used OpenNeuro ds004473 version 1.0.2; the source metadata and cached version-pinned record identify this exact release as CC0.
- doi:10.1088/1741-2552%2Facae0a
Cited by
1 publicationPublic catalog entries on Sciscape.
- supersedesMatched-trial fitting gave no clear gain in later single-window Brier error in eight SEEG recordings after 32 early trials
Package version 2 fixes the file checker for the supplemental metadata folder added by sci fetch. The analysis, data, results, and figures are unchanged.