Three forward validation blocks did not show a clear Brier error reduction for separate power and phase penalties in eight fixed SEEG records
We tested whether three earlier validation blocks improve a SEEG reader compared with one block. The reader uses separate penalties for power and phase features. Inputs and later test trials were fixed in eight public OpenNeuro ds004473 version 1.0.2 records. The mean later-trial Brier error was 0.181806 with one block and 0.181137 with three blocks. The paired change was -0.000669, with a 95% whole-person bootstrap interval of [-0.002016, +0.000143]. This result does not show a clear error reduction or equal performance. A planned rectangular-taper check gave an uncertain change in the opposite mean direction. The study is exploratory because the same people and test trials were used before. Contact selection was fixed and response times were supplied. The package preserves exact selected signal samples, source records, every fitted model, predictions, three figures, code and offline repeat instructions. All 64 prior one-block metric rows matched exactly. Independent signal and model checks passed. A full repeat gave identical hashes for all 24 core analysis files.
stereoelectroencephalography · brain-computer interfaces · forward validation · Brier error · power and phase features · regularization · exploratory study
References
5- Separate power and phase penalties reduced mean Brier error against a shared penalty on later trials in eight selected SEEG records sci:sha256:52c83dfcc8e81bd04f93edff10c72e5c45f7632911769c2f4448175b2904737c
[Legacy relation: reproduces] Use the separate penalty engine, candidate grid and one-block rule. All 64 archived one-block metric rows and selected C values match exactly across both tapers. The new test changes the validation schedule and keeps these inputs fixed.
- usesAdded imaginary coherency increased mean Brier error for tuned linear readers in eight SEEG records with fixed 500-sample inputs sci:sha256:2263fe7cecddd0d4616d9f4cfb58a07f322990304de5c63f2a497260c14e6bb5
Use the exact selected bipolar samples, spectral feature code, pinned source metadata and source range recipe. These inputs fix the signal representation for the new validation schedule test.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
Keep the original contact selection, retained trials, exclusions and chronological outer split. The inherited selection records and publication manifest are preserved.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The fixed input samples and source metadata are from OpenNeuro ds004473 version 1.0.2.
- usesdoi:10.1088/1741-2552%2Facae0a
This paper reports the source SEEG recordings used by the dataset.
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesPre-cue fixed 20 Hz phase did not show a clear response-time error decrease beyond power and history in eight SEEG test blocks
Used its phase task-state result as context to define a different response-time question. Its data arrays and code were not used.