A new fit did not show a clear gain in 50 ms offline theta phase forecasts after stimulation in seven paired SEEG rest records
We tested whether a new fit after stimulation reduced theta phase forecast error. We used paired public SEEG rest records from seven people in OpenNeuro ds006065 version 1.0.0. We kept four pairs per person, selected from early rest before stimulation. Models used only past input samples. We scored later data that were not used for fitting. At a 50 ms delay and with a 2 s centered offline phase reference, the fixed fit gave 57.45 degrees mean error after stimulation. The new fit gave 57.81 degrees. The mean gain from the new fit was -0.36 degrees; the participant bootstrap 95% interval was -1.22 to 0.58 degrees. The exact paired sign-flip test gave p = 0.46875. Only two of seven people had a lower error with the new fit. All nine fixed reference and delay checks had intervals that included zero. The result does not prove equal performance. The selected records contain four phase-locked sessions and three control sessions. They cannot show a causal effect of stimulation. The phase target uses future samples, so this is an offline forecast benchmark. The package includes voltage inputs, code, source hashes, models, tables, figures, and replay instructions.
SEEG · theta phase · phase forecasting · ridge regression · brain-computer interface · offline phase reference · model transfer · calibration
References
3- usesA learned linear filter reduced mean 50 ms offline theta phase error by 12.54 degrees in seven SEEG rest recordings sci:sha256:681793db790925fefcdd246f20a125cf7ec0a4d0c7ac415b89cd84c8e64b5d35
Supplied the exact pre-rest voltage, fixed four-pair choice, offline phase reference, and ridge fit functions. We added snapshot-checked post-rest voltage and tested forecast transfer between the paired records.
- usesdoi:10.18112/openneuro.ds006065.v1.0.0
Source of the pre-rest and post-rest SEEG records and recording metadata; the exact analyzed snapshot is version 1.0.0.
- doi:10.1038/s41467-025-59417-7
Describes the source study of phase-locked stimulation in these records; this publication tests forecast transfer and does not reproduce its connectivity analysis.
Cited by
4 publicationsPublic catalog entries on Sciscape.
- usesFour pre-rest confidence bins reduced 90 percent target arc scores for 50 ms offline theta forecasts on post-rest SEEG in seven people
This paired-rest study supplied all 14 retained voltage files, 28 fixed pre-rest models, source metadata, and reference code. These inputs enabled the test of frozen confidence limits on later post-rest without refitting forecasts.
- usesUpdates using 56 seconds of past labels gave no clear gain in 50 ms offline theta phase forecasts in seven SEEG rest records after stimulation
We reuse all 14 compact pre-rest and post-rest voltage records, the fixed four-pair lists, source metadata and hashes, and the ridge and phase-reference functions. We extend the one-fit test to repeated fits with strictly delayed labels. Our frozen pre-rest weights match that publication exactly.
- usesA 56 second threshold reset showed no clear gain in offline command retention after stimulation in seven SEEG record pairs
We used 14 paired voltage inputs, source metadata, and raw extraction receipts for seven people. We compared gates on the same later post-rest data.
- usesA 56 second threshold reset showed no clear gain in offline command retention after stimulation in seven SEEG record pairs
We used 14 paired voltage inputs, source metadata, and raw extraction receipts for seven people. We compared gates on the same later post-rest data.