In 8 SEEG patients, 6–15 Hz pre-stimulus magnitude weakly predicted CCEP N1, but magnitude-weighted phase did not

We tested whether weighting pre-stimulus phase by oscillation magnitude reveals a phase effect on single-trial CCEP N1 amplitude, and whether pre-stimulus magnitude itself predicts N1. We reused the archived single-trial table from 8 patients (OpenNeuro ds005448 v1.0.0; 1–2 mA clinical single pulses; 1,078 response-site units with at least 6 trials). Phase and magnitude were estimated with a 3-cycle wavelet ending 10 ms before each pulse at 4, 6, 10, 15, and 20 Hz. Within each unit, phase statistics used either equal weights or magnitude weights; the phase-only permutation null preserved the pairing of N1 and magnitude. We also measured the correlation between N1 and log magnitude. Group inference used per-patient permutation z scores and an exact two-sided sign-flip test over 8 patients. Magnitude-weighted phase did not predict N1 (all real-pulse p ≥ 0.12), and real-minus-sham phase tests were not significant. Sham phase statistics were as large as or larger than the real-pulse values. In contrast, pre-stimulus magnitude at 6–15 Hz had a small positive correlation with N1 (mean r = 0.047–0.055); at 10 Hz, r = 0.055, all 8 patient correlations were positive, and p = 0.0078, below the Bonferroni threshold of 0.01 for five power tests. Sham correlations were near zero, and real-minus-sham p = 0.016 at 6 and 10 Hz. An r near 0.05 corresponds to about 0.3% of single-trial variance, suggesting only a small gain for a state-triggered stimulation approach. This is not a fully independent replication of related work because the analyses used the same patients and runs. Other limits include 100 permutations per patient, the short wavelet estimate, response-site measurements only, clinical pulses only, no rejection of interictal discharges, and about 10 pulses per unit. A longer or causal-forecast estimator and stimulation-site state were not tested.

CCEP · SEEG · N1 amplitude · single-trial analysis · pre-stimulus state · wavelet magnitude · phase · permutation test · brain-computer interface

Admitted 2026-11-09 18:48 UTC5 citing publications

References

supportsPre-pulse 4–20 Hz phase did not predict CCEP N1 amplitude in 8 SEEG patients sci:sha256:65c60c58375bbd719b8dc38c3d1e55ff74b64e402df147586b75a6717f8a37f1
Magnitude-weighted phase tests with a phase-only permutation null also found no phase effect on the N1. Thus the null is not caused only by noisy phase in weak-oscillation trials.
usesPre-pulse 4–20 Hz phase did not predict CCEP N1 amplitude in 8 SEEG patients sci:sha256:65c60c58375bbd719b8dc38c3d1e55ff74b64e402df147586b75a6717f8a37f1
The input is the archived single-trial phase and magnitude table of this publication, used without change (SHA256 in provenance).
supportsPre-stimulus 4–30 Hz power weakly tracked single-trial CCEP amplitude in eight SEEG patients sci:sha256:6badf6d376d57ff26ae9bbad5a43b4f303f32605d250ca6ec84a021a5c5e4579
On a different random pair sample and with a different power estimate, pre-stimulus 6-15 Hz magnitude at the recording channel again gave a small positive correlation with N1 (r about 0.05), and the sham r was near 0.
usesdoi:10.18112/openneuro.ds005448.v1.0.0
The analyzed archived single-trial table derives from OpenNeuro ds005448 v1.0.0, as recorded in provenance/source.json. The exact-version dataset metadata records CC0 and this DOI.

Cited by

5 publications

Public catalog entries on Sciscape.

challengesThe 10 Hz pre-stimulus magnitude effect did not replicate in 8 Mayo SEEG patients
Same pre-registered pipeline in an independent center (ds004696, 6 mA): the 10 Hz pre-stimulus magnitude effect on single-trial N1 did not replicate (r -0.022, 2/8 positive, p 0.98; pooled CI excludes 0.055).
supportsA pre-pulse spectral-state model did not choose larger CCEP N1 pulses better than a sham control
A small positive real-pulse gain agrees with the small 6-15 Hz magnitude effect; the sham control shows that about half of a combined-feature gain also occurs without a pulse.
usesDistant SEEG CCEP variation was not greater than sham in 7 of 8 patients
Retrieval copy of the same single-trial table (same SHA-256); its pre-stimulus state result is the per-site context for this shared-variation test.
usesThe early CCEP N1 did not decrease across five pulses at 5 s intervals in eight SEEG patients
Copy of the same single-trial table; checksum matches.
challengesPre-pulse 10 Hz magnitude did not predict CCEP N1 in 13 patients with 1 Hz SEEG trains
We tested its claim (higher pre-stimulus 6-15 Hz magnitude gives a larger single-trial N1, 10 Hz r 0.055) at a third center with the same measures. In ds006254 the 10 Hz pooled r was -0.003 (95% CI -0.024 to +0.019, which excludes 0.055), 7/13 patients positive, p 0.55.