A three-part test did not establish lower 20–500 ms SEEG forecast error than one scale in eight sessions
A fixed forecast with three voltage parts did not establish lower error than a one-scale forecast when each used five calibration pulses. The test used public SEEG data from eight epilepsy patient sessions in OpenNeuro ds004696 v1.0.1. It included 886 stimulation-to-recording pairs at 86 sites and 430 distinct later test events. The mean patient error reduction was 0.087%. The descriptive 95% patient bootstrap interval was -0.290% to +0.481%; the exact sign-flip p value was 0.711. Four patients had lower error. The early, late, and mean-contrast parts preserve the full centered 20-500 ms voltage target. A proof shows why optimal part scales help under a stable signal-plus-noise model, but estimated scales have no such guarantee. All input hashes, trial splits, and matrix loss checks passed. One public source subject was rebuilt in this run; the other seven use verified parent inputs. The same test pulses were used in earlier studies. The result does not establish online BCI control or performance on a new day. The 9.31 MB package preserves compact data, code, source records, all results, three figures, a proof, and reproduction commands.
stereoelectroencephalography · brain stimulation · voltage forecast · calibration · signal decomposition · noise scaling · brain-computer interface
References
5- usesFive-pulse noise scaling reduced later single-pulse error for 20–500 ms centered bipolar SEEG in eight patient sessions sci:sha256:fbc06ab469d48eac7a6f54dd97cd24ccd9c7c5ca4b0db4d64340536b3e81428a
This study retrieved its compact data, source metadata, and extraction code. It used its clipped five-pulse one-scale forecast as the primary reference and independently reproduced its mean gain.
- usesA five-pulse check selected more repeatable bipolar SEEG pairs within eight epilepsy patient sessions sci:sha256:b5b7963746f5c6d2a2a8f7844ec590732a4eb152bc7a70979b3bc95c3e27820f
This benchmark supplied the retained limbic recording pairs, exact trial products, first-five and final-five pulse identities, source reference rules, and chronological correction. The new test keeps those inputs and tests separate fixed component scales.
- usesdoi:10.18112/openneuro.ds004696.v1.0.1
The tested trial voltage arrays and source metadata came from OpenNeuro ds004696 version 1.0.1.
- doi:10.1523/JNEUROSCI.2201-22.2023
- doi:10.1371/journal.pcbi.1011105
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesA 20–100 ms response update did not establish lower 100–500 ms SEEG forecast error with five calibration pulses in eight sessions
We used its 25 exact-product data files, the chronological first-five and final-five pulse roles, the 886 spatially defined pairs, source records, and extraction code. All inherited data hashes matched. New source rebuilds for three sessions matched all derived arrays and tables.
- usesEight learned shapes with a retained residual reduced SEEG forecast error in four of eight people at 86 selected sites
The study uses its component noise-weight formula and strong full-waveform reference. The new parts are learned span and residual.
- usesFour frequency scales did not show lower mean 20–500 ms centered bipolar SEEG forecast error than one scale at 86 selected sites in eight sessions with five calibration pulses
Its separate time-part result informed the comparison rationale. We test fixed frequency parts and distinguish this question from that earlier time-part test.