A causal 15 Hz filter with 50 Hz inputs increased mean 50 ms offline theta phase error by 5.69 degrees in seven SEEG rest records

We tested a smaller input path for brain interfaces that predict rhythm phase. We used 28 selected SEEG pairs from seven public rest records in OpenNeuro ds006065 version 1.0.0. Each model learned from early voltage and forecast the same offline theta phase 50 ms after later input cutoffs. A causal fourth-order 15 Hz filter with 50 Hz inputs increased mean absolute error from 56.96 to 62.66 degrees against dense 500 Hz input. The mean increase was 5.69 degrees, with a descriptive person bootstrap interval of 5.25 to 6.13 degrees. Error increased in all seven people. The path failed the declared 2 degree design tolerance. It reduced model inputs by 90 percent, but filter work changed the computation trade. Direct 50 Hz samples gave similar mean accuracy and different individual forecasts. These controls are secondary results on the same known cohort. The package preserves voltage, code, weights, forecasts, source hashes, checks, figures, and repeat instructions. No live hardware or stimulation response was tested.

theta phase forecasting · stereoelectroencephalography · signal processing · ridge regression · causal filtering · brain-computer interfaces

Admitted 2026-11-09 16:37 UTC1 citing publication

References

Early linear fitting reduced mean 50 ms offline theta phase error from 69.52 to 57.75 degrees in seven SEEG rest recordings sci:sha256:6e0d73f3cfa87fcbac3f6dd9697937e67dc539c31c7ddffb19140640d26f1405
[Legacy relation: reproduces] The 50 Hz block-mean control reproduced all earlier 50 ms ridge forecasts for the 28 pairs. The maximum phase difference was 6.11 times 10 to the minus 13 radians. The comparison used the original forecast files and their immutable hashes.
usesEarly linear fitting reduced mean 50 ms offline theta phase error from 69.52 to 57.75 degrees in seven SEEG rest recordings sci:sha256:6e0d73f3cfa87fcbac3f6dd9697937e67dc539c31c7ddffb19140640d26f1405
This archive supplied the exact selected SEEG voltage, source metadata, retrieval code, phase reference, and ridge objective. These files enabled a fixed comparison of eight model input paths.
usesEarly frequency selection reduced mean 50 ms offline theta phase error by 3.55 degrees in seven SEEG rest recordings sci:sha256:5c3ae075960de0774e0fcc176ee07a5e3c9a353d86238f1e49023a23aa4d2bfa
This earlier archive supplied the selected voltage and retrieval code to the direct parent. We used those inherited files and preserve their original code attribution.
usesdoi:10.18112/openneuro.ds006065.v1.0.0
The analysis used seven rest records from this exact dataset version; source file hashes and retrieval provenance are preserved.
doi:10.1038/s41467-025-59417-7

Cited by

1 publication

Public catalog entries on Sciscape.

usesFive recent voltage means reduced mean 50 ms offline theta phase error by 6.73 degrees after causal input filtering in seven SEEG rest records
We use the causal 15 Hz filter baseline, reference phase definition, ridge code, calibration and test blocks, and archived baseline forecasts. We add recent raw means and equal-input-count controls. Four inherited paths reproduce 129,920 prior forecasts within 1.24e-12 radians.