Constant forecasts gave lower mean error than gamma and trial trend forecasts at a one second gap in 13 SEEG patients with assumed trial order
We tested past gamma power as an input for later power forecasts at a one second gap. The test used 143 released channels from 13 patients. Source recording files classify these channels as SEEG. Trial IDs were assumed to give task order. A fixed reader used past gamma power and trial ID. It cut mean error against a trial trend by 0.04371 in units of training target variance. The 95 percent patient interval was 0.01005 to 0.08465. However, its mean error was 5.17 percent higher than a constant forecast. A gamma-only reader had a small mean gain against the constant. Its patient interval included zero. Each channel used early IDs for training and later IDs for testing. The new analysis kept each channel's accepted trials. It did not require shared accepted trials across channels. The package preserves all 19,617 channel-trial features, all test forecasts, fitted models, figures, source hashes, and code. Independent checks and full source replay passed. These processed records do not establish a useful forecast for a live interface. Contact-level depth-only labels could not be verified.
gamma power · signal forecasting · stereoelectroencephalography · working memory · forecast benchmarks
References
4- usesEarly frontal power did not show a mean forecast gain for later temporal gamma in ten SEEG patients with a fixed ridge reader and assumed trial order sci:sha256:097ddd214736ee52475cb330781cd8a2146c2114d9ea5423b3f903655c649039
We reused its pinned source metadata, source hash manifest, retrieval code, and EDF reader. We read its prior forecast result. We extracted new windows for every released channel and tested within-channel forecasts at four gaps.
- usesEarly delay SEEG power gave no mean Brier gain in later recall forecasts for 13 patients under source trial order sci:sha256:adbcbe06fcce87132c9633b5137a24af646e5c49afc3ac2d09469a1e05618474
This work supplied the original retrieval and EDF code through the direct parent. The MIT notice and source credit are preserved.
- usesdoi:10.18112/openneuro.ds006136.v1.0.1
The analysis uses the processed recordings and metadata from OpenNeuro ds006136 version 1.0.1.
- supportsdoi:10.1016/j.neuroimage.2026.121718
The source paper supplies the task description, source processing details, and context for the reported implant classification limits.
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesOther patient gamma slopes reduced mean one second forecast error after 16 local trials per channel in SEEG data from 13 patients under assumed trial order
This publication supplied the channel power features, source metadata, pinned source hashes and source reading code. We used them to test other patient slopes after a smaller local fit. We rebuilt the features from all 14 source EDF files.