A fixed pre-pulse voltage model had a mean 3.8% error increase at later sites in five SEEG sessions

We tested a fixed model on 376 pulses at 94 later stimulation sites in five public SEEG sessions. Each site used four earlier pulses for calibration. All common model coefficients came from earlier sites in the same patient. The added input was pre-pulse voltage size. Its mean relative error increase was 3.8% against a calibration and pulse-order baseline. Two patients improved and three worsened. A descriptive patient bootstrap interval for prediction gain was -11.9% to +1.7%. The result does not establish a population effect. The package supplies a strict time split, source voltage bins, all predictions, three figures, and code. All 110 pinned source files passed hash checks. Fresh raw extraction matched all 30 array fields and five event tables. Independent formula checks, repeat runs, and test target isolation checks passed. The target contains background activity and possible residual artifact. This study does not establish a causal brain state effect or a live stimulation policy.

stereoelectroencephalography · brain-computer interface · single-pulse stimulation · response prediction · pre-stimulus voltage · time-split validation · root mean square voltage

Admitted 2026-11-09 18:51 UTC4 citing publications

References

usesA fixed SEEG distance rule accepted most pulses from unlearned stimulation shafts in five sessions sci:sha256:ecfcd951080fd4edc42a41327003efe86593c463b7a6afdbbcad8732b89e3028
We reused its source voltage bins, BIDS metadata, fixed recording contacts, and pinned source manifest. We verified their hashes and rebuilt the bins from raw files. We did not reuse its fitted response recognition models.
usesRemoving missing features raised mean site accuracy for 20 to 100 ms responses under known simulated contact loss in five SEEG sessions sci:sha256:ee49903b43a054e26fbb0d85370a8cc3bbeb395ca7baf6c0069fcd7329eac507
The inherited raw extraction recipe and contact configuration trace to this source-verified dataset package. Its extraction and retrieval code was preserved through the immediate input publication and used for an exact fresh reconstruction.
usesdoi:10.18112/openneuro.ds004457.v1.0.2
The analysis used data from OpenNeuro ds004457, version 1.0.2; the pinned dataset metadata records this exact version and its CC0 license.
doi:10.1523/JNEUROSCI.1325-22.2023
doi:10.1016/j.brs.2026.103118

Cited by

4 publications

Public catalog entries on Sciscape.

usesFour signed pre-pulse modes per pair raised mean relative 20–100 ms SEEG forecast error by 19.2% at 94 later sites in five sessions with ridge strength 0.1
Used its quality loader and strict earlier-to-later site split. The prior size forecast motivated this separate signed waveform test. The complete site split matched its saved split.
usesA four-mode background model had a mean 0.62% increase in 20 to 480 ms SEEG template error after contact order reversal in eight STReEF patients at ridge strength 0.1
Its different-dataset result informed the matched background control. No code or data from this publication entered the present scores.
usesA fixed pre-pulse adjustment gave 70.8% accuracy versus 71.4% without adjustment for neighboring sites at 20–100 ms in five offline SEEG sessions
We read its pre-pulse response-size forecast results after our primary scores. We used them to explain the different scope of our waveform test.
usesBackground voltage prediction reduced mean centered SEEG error at 20–100 ms by 3.89 percent compared with four-pulse templates at 94 later sites in five sessions
Used its source quality loader and strict earlier-to-later site split through the immediate source. The saved split matched the full split in this study. We retained the main loader and documented its stimulation-contact status limit.