Recent pulse templates reduced error by a mean of 2.51% relative to initial templates at 3 to 15 ms in 20 SEEG records

We tested whether a recent voltage template gives less late error than an initial template during SEEG stimulation. We used the first record from each of 20 people in OpenNeuro ds005169 version 1.0.0. Both fits used 24 pulses. The separate test used 24 late pulses and 1671 good channels off the stimulation shaft. At 3 to 15 ms, the recent template reduced error by a mean of 2.51% relative to the initial template. The whole-person 95% bootstrap interval was 0.08% to 5.58%. 13 of 20 people had lower error. The exact sign-flip reference gave p = 0.075. The recent template did not clearly reduce error relative to uncorrected voltage. Some detected pulse times were uncertain. These data show a small observed gain over an older fit, with weak evidence for a general benefit. The score does not test neural recovery or BCI decoding. The package includes new late-pulse products, exact score replay, raw source hashes, code, figures, and timing controls.

SEEG · intracranial EEG · electrical stimulation · pulse template · voltage prediction · brain-computer interface

Admitted 2026-11-09 18:25 UTC1 citing publication

References

usesOne early voltage scalar increased prediction error by a mean of 0.31% at 3–15 ms in 20 SEEG trains after 24 fit pulses sci:sha256:267aaac0e6978135ad050c3c51f1d4af411cf65ec8ec2859d4084d03973c3559
This publication supplied the inspected raw source pins, 120 sidecar files, and extraction code. We adapted its header parsing and median edge detector. We extracted new late pulse blocks from all 20 hash-checked raw EEG records. No saved voltage array or error score from that publication was used.
usesRepeatable pulse class differences exceeded baseline at 3 to 15 ms after detected edges in 17 of 20 SEEG recordings sci:sha256:879b5223cbf14d14e8e6ed3642b3968e076bf8d799c75c3c86fea26e328c4ab9
This is the upstream extraction work identified in the reused source recipe. Its pulse-class and raw-source extraction method informed the new equal-fit-count test on later blocks. The present package credits that indirect code dependency.
usesdoi:10.18112/openneuro.ds005169.v1.0.0
The analyzed raw SEEG source is OpenNeuro ds005169 version 1.0.0; its saved dataset metadata records this version-specific DOI and CC0 license.
doi:10.1002/hbm.25749

Cited by

1 publication

Public catalog entries on Sciscape.

supportsTen added pulses reduced stored SEEG response error at 50 to 250 ms in nine file-order tests
That report showed why a gain over an initial SEEG pulse template requires a separate uncorrected-voltage reference. Our independent 50-to-250-ms test supports this limitation: the update beat the fixed map in nine sessions but a zero map in only five. We inspected the earlier report after our scores and reused no code or data.