Linear filters fitted on early data reduced 50 ms offline theta phase error against fitted oscillators in seven SEEG rest recordings
Linear filters fitted on early SEEG data gave lower later reference phase error than calibrated oscillators. The test used 28 selected bipolar contact pairs from seven people in OpenNeuro ds006065 version 1.0.0. Each forecast used 0.5 seconds of past voltage. Settings were selected on an early validation block. At a 50 ms delay, mean absolute phase error fell from 69.52 to 58.22 degrees. All seven people and all 28 pairs had lower error. The reference was a two-sided filter in the nominal 4 to 10 Hz band. The mean reduction was 11.30 degrees. The descriptive person bootstrap interval was 9.85 to 12.85 degrees. A fixed filter with matched mean and trend removal gave 60.53 degrees in a post hoc check. The extra reduction against this control was 2.31 degrees. A causal output norm gate retained 51.52 percent of forecasts and gave 46.86 degrees of error. All full source hashes and extracted voltage values were checked. The package preserves inputs, code, models, forecasts, figures, and reproduction instructions. The cohort was reused. This is an exploratory rest signal result. No stimulation effect was tested.
SEEG · phase forecasting · theta phase · linear filter · ridge regression · brain-computer interface · offline phase reference · intracranial EEG · forecast error
References
5- usesEarly frequency selection reduced mean 50 ms offline theta phase error by 3.55 degrees in seven SEEG rest recordings sci:sha256:5c3ae075960de0774e0fcc176ee07a5e3c9a353d86238f1e49023a23aa4d2bfa
Used the exact 28 selected bipolar voltage traces, early frequencies, source metadata, and retrieval code. Reproduced the original oscillator forecasts. Checked all extracted voltage values against fresh full source downloads. Added early fitted linear filters and a separate phase correction for each control delay.
- usesdoi:10.18112/openneuro.ds006065.v1.0.0
The study used OpenNeuro ds006065 version 1.0.0; the recorded version tag and source file hashes identify the analyzed snapshot.
- doi:10.1038/s41467-025-59417-7
- doi:10.7554/eLife.68803
- doi:10.1016/j.neuroimage.2020.116761
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesBroad and early-peak theta references differed by more than 90 degrees at 14.1 percent of later times in seven SEEG rest records
We preserved its full source hash record as earlier provenance evidence. We also ran new source checks. This record does not add an independent cohort.