Power blocks with 50 percent overlap increased correct SEEG outputs with a guarantee for a fault of ten samples on one contact in eight recordings at C = 1
This study compared two fixed task-state readers on public SEEG data from eight recordings. Both used the same 500-sample windows and 24 fitted features. The primary model used C = 1. Nine power blocks with 50 percent overlap increased the equal-person fraction of correct outputs with a score-sign guarantee from 5.48 to 10.23 percent. The increase was 4.75 percentage points. The descriptive paired 95 percent person bootstrap interval was 2.14 to 8.21 points. All eight recordings showed an increase. The guarantee permits any finite added waveform on one source contact within at most ten consecutive samples. A proof gives the bound and its use on a current input. Exact replay, separate model fits, waveform tests, and 16 public source range checks passed. Most windows had no guarantee. An output with a guarantee can still have the wrong label. Overlap gave lower correct guaranteed fractions for one-sample and 100-sample faults. This exploratory result uses a reused clinical cohort and offline labels. It does not test stimulation or live use. The archive keeps exact inputs, code, proof, figures, results, source records, and repeat instructions.
stereoelectroencephalography · SEEG · brain-computer interface · signal processing · overlapping power blocks · fault tolerance · median power · robust classification · epilepsy monitoring · exploratory analysis
References
4- usesTen spaced gaps gave lower mean zero-fill SEEG accuracy than a circular gap block at 20 percent simulated sample loss in eight fixed decoders sci:sha256:35c3a178c2212bd9b314433661dfc484678cb20f0c291b7c94bc044397f7c9b9
This study uses native integer 500-sample SEEG windows and pinned public source retrieval records from this work through the direct parent. Fresh source range checks verify 16 trial ranges.
- usesMedian block powers gave higher task-state accuracy than mean powers for simulated 1 mV, 10-sample SEEG faults on one contact in eight recordings sci:sha256:7e4be46c9c145b5de38d258bb39ea502a49c77fca52c1e1d18681de88d768b0a
This study uses the five-block median power method, its one-contact fault bound, its exact selected inputs, and its fixed trial and pair selection. The five-block primary result was reproduced before the overlap comparison.
- 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 study holds the inherited early pair ranking, event state definitions, and trial exclusion rules fixed. The archive preserves the original selection code for inspection. It does not repeat the selection.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The study uses retained SEEG windows from OpenNeuro ds004473 version 1.0.2; source URLs, pinned object versions, hashes, and range checks are preserved in the package.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesSine fits did not establish lower Brier error than linear repair in eight SEEG records with matched training and 20 percent periodic loss
This study supplies the exact integer SEEG windows, source recipes, and fixed selected pairs. We keep its trial order and split while testing two power maps under periodic gaps.
- usesSine fits did not establish lower Brier error than linear repair in eight SEEG records with matched training and 20 percent periodic loss
This study supplies the exact integer SEEG windows, source recipes, and fixed selected pairs. We keep its trial order and split while testing two power maps under periodic gaps.