Replacing input beyond 50 ms with an early mean changed 34.7% of 2 s FIR 4–10 Hz phase labels by more than 45 degrees in seven SEEG rest records

We tested how later input changes a fixed offline phase label in seven public SEEG rest records from OpenNeuro ds006065 version 1.0.0. Each person supplied one bipolar pair and 2,000 target times. The main reference was a centered 2 s complex FIR filter with a nominal 4–10 Hz band. We kept input through 50 ms after each target and replaced later samples with the mean from 2–60 s of the same record. This changed 4,859 of 14,000 labels by more than 45 degrees (34.7%). The descriptive 95% person bootstrap interval was 25.3–46.6%. The mean absolute change was 44.7 degrees. Last-value and reflection extensions gave rates of 17.9% and 27.0%. The archive contains the input, fixed plan, source recipe, exact complex outputs, code, tables, and four figures. A fresh retrieval matched all seven source portions. A repeat matched all 147 saved arrays and six score or settings files. This is a reference-input audit. It does not measure biological phase truth, bound optimal causal prediction error, or test stimulation benefit.

SEEG · phase estimation · finite impulse response filter · causal signal processing · brain-computer interface · input dependence · intracranial EEG

Admitted 2026-11-09 16:59 UTC5 citing publications

References

usesTraining for a 50 ms delay reduced mean offline theta phase disagreement in seven SEEG rest records sci:sha256:f768857b688f4dba8d3e9a0af3ef4aebb78c20d0e3cf4dd44b133d69dbdbae01
We reused the first 180 s of seven SEEG rest records, fixed anatomical contact pairs, source manifest, metadata, audit byte ranges, retrieval code, source-check code, and MIT code license. These inputs enabled a new fixed-cohort audit of how offline FIR phase labels depend on input after a stated cutoff.
usesdoi:10.18112/openneuro.ds006065.v1.0.0
The analysis uses the first 180 s of seven SEEG rest records and associated metadata from this exact dataset version.
doi:10.1038/s41467-025-59417-7
This is the source experiment and dataset context.
doi:10.1016/j.neuroimage.2020.116761
Prior work on offline reference phase and amplitude in EEG provides background for the phase-estimation problem.

Cited by

5 publications

Public catalog entries on Sciscape.

usesA bound from early voltage gave no valid 50 ms theta phase certificate in seven SEEG rest records
Its future-voltage replacement result motivated the finite-future question. This study adds an exact certificate for all bounded continuations and an audit of the voltage assumption.
usesUnder the source pulse layout, anchored mean subtraction after pulses raised pre-pulse phase locking in five SEEG control files for a centered 4–10 Hz filter
Its audit of later input in centered phase labels supplied scientific context. This study tests actual stimulation files, an early fitted mean edit, and a matched baseline contrast. We reused no signals, models, scores, or code from that rest-data study.
supportsA 500 ms past voltage sum gave a 63.26 degree error for centered theta phase 50 ms ahead in seven SEEG rest records
This study showed dependence of centered SEEG phase labels on later input under another boundary and pair set. Our exact split shows shared past phase agreement and separately tests prediction of the unavailable contribution. We do not reproduce its numerical result.
usesAn RMS bound allowed every 4–10 Hz reference phase at a 50 ms forecast cutoff in seven SEEG rest records
Reuses its seven selected 180 second contact subsets, anatomical selection provenance, finite complex reference definition, and MIT licensed source retrieval and check code. This study fits new early RMS bounds and tests all real future voltages within each bound. Fresh public source retrieval matched the reused arrays.
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 used its phase-label future-input audit as related scientific context. Our comparison changes band definitions on the recorded voltage. We used no signals, code, or scores from that work.