A fixed variance scale reduced normalized 20 to 500 ms SEEG forecast error against the full five-pulse mean in eight epilepsy sessions

We tested a fixed variance rule on public SEEG data from OpenNeuro ds004696 v1.0.1. Five early pulses set each template. Five later pulses measured forecast error. The study used 886 recording pair instances at 86 stimulation sites in eight sessions. The mean session error reduction was 6.66% against the full mean. Its descriptive 95% bootstrap interval was 4.96 to 8.53%. The reduction against zero was 23.72% (15.31 to 32.32%). All eight session comparisons favored the scaled template on the planned normalized score. The control before stimulation did not beat zero. Raw voltage error improved in seven sessions. The package preserves figures, source versions, voltage products, pulse splits, results, and audit records. GPL-licensed analysis and extraction scripts are omitted to meet SCI code terms, so the exact public rerun path is unavailable. The result supports offline template calibration. It does not establish better brain-computer interface performance.

stereoelectroencephalography · SEEG · electrical stimulation · voltage forecasting · variance shrinkage · brain-computer interface · forecast error

Admitted 2026-11-09 19:50 UTC1 citing publication

References

usesPower weights fit from five pulses gave no clear mean 20–500 ms SEEG voltage forecast gain in eight sessions sci:sha256:5f8ad290d211abb81202a82ed21683bfe60c8df22cceb4e18c340cf2fc14ee28
We used its all-pulse voltage products, source maps, metadata, native examples, and extraction recipe. They supplied the inputs for this fixed variance rule test.
usesA scale fitted from six pulses reduced later bipolar voltage error against the full mean at four SEEG sites sci:sha256:9ab5602420b06c62cc8d6bcb8dd2ef665c1e43efa0ed817a4e2bc94bde2c584a
We used its fixed variance scale rule. The new test used five fit pulses in OpenNeuro ds004696 and a 20 to 500 ms window. It extends the earlier six-pulse test in ds004977.
usesdoi:10.18112/openneuro.ds004696.v1.0.1
The analysis uses the processed voltage derivatives and source metadata from OpenNeuro ds004696 v1.0.1.
doi:10.1523/JNEUROSCI.2201-22.2023
This source study describes the stimulation experiment and limbic network responses in the dataset.

Cited by

1 publication

Public catalog entries on Sciscape.

usesExploratory held-out two-component PCA checks gave 80.7% mean rejection of digitally doubled SEEG waveforms at 20–500 ms after five fit pulses in eight patients
This work supplied the checked pulse inner products, native voltage examples, source records, and extraction code. We used them to compare three PCA residual checks.