Bounded baseline variance weights raised mean relative 50 to 250 ms SEEG prediction error by 1.07 percent in five patients with separate groups of five pulses
We tested bounded baseline variance weights for a SEEG waveform estimate made from five pulses. The source was OpenNeuro ds004457 version 1.0.2. The sample had five patients and 1080 selected links. The test target was the mean of five separate pulses. We tested all 252 groups of five from ten at 50 to 250 ms. The equal mean patient error reduction was -1.07 percent. Thus mean relative error rose by 1.07 percent in this finite sample. Two patients improved and three had higher error. Descriptive patient resample limits were -6.95 to 6.19 percent for error reduction. The exact two sided patient sign flip p value was 0.875 under independent patients and null sign symmetry. The same weights reduced before-stimulation error by 13.55 percent. That result did not transfer to a mean after-stimulation benefit. This test does not establish equivalence, a population harm, or clinical value. The package preserves inputs, all split errors and weights, source records, code, figures, and repeat instructions. A full raw extraction matched all 3477760 saved voltage values. A complete repeat gave identical bytes for all 14 result files.
stereoelectroencephalography · SEEG · pulse averaging · variance weighting · waveform prediction
References
4- usesBootstrap width ranked large peak shifts in five-patient SEEG data (mean AUC 0.667) sci:sha256:fd99edb6aa2bc60793962830f8b0939fc405d8ab1a3345e552af479ccaca828d
Used its saved pulse voltage arrays, BIDS metadata, extraction code, retrieval code, and source locks. These files supplied the measured SEEG inputs and a complete raw rebuild.
- usesA variance rule reduced average 50 to 250 ms SEEG prediction error in five patients with separate five pulse groups sci:sha256:335cdede136528c2246142ff023095aeaebe4b09b0b78a45abc1cd9dc2cc0387
Used the fixed scalar variance shrinkage rule as a comparator and its separate pulse prediction framework. The current constant-window exclusion changes the comparison sample.
- usesdoi:10.18112/openneuro.ds004457.v1.0.2
The analysis used the exact OpenNeuro ds004457 version 1.0.2 source, identified by the saved source lock and version-specific dataset metadata.
- doi:10.1523/JNEUROSCI.1325-22.2023
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesTrimming five training pulses reduced 50 to 250 ms SEEG prediction error in two of five selected patients
We used its individual pulse arrays, source locks, exclusion rules, baseline and score definition, and raw rebuild code. We matched its ordinary and shrink errors exactly. The trimming test is new in this fixed pulse pool.
- usesMedian estimates increased mean relative 50 to 250 ms SEEG prediction error by 10.45 percent for separate five-pulse groups in five patients
We reused its five pulse-window arrays, fixed contact and event selection, source metadata, and pinned retrieval and extraction code. These inputs enabled the new median comparison. We retrieved the public raw files and rebuilt every saved array. This study uses the same patients and is not an independent patient replication.