A graph correction reduced reverse voltage size error in six offline SEEG maps at 20 to 500 ms

We tested whether other contact pairs could correct a reverse voltage size forecast. The data were public single-pulse SEEG recordings from OpenNeuro ds004696, version 1.0.1. Each tested pair was absent from its graph fit. Separate final pulses supplied the outcome. The test included 167 pairs in six selected sessions. For 20 to 500 ms, equal-patient mean squared log error fell from 1.943 to 0.741. This was a 61.8 percent reduction. Mean error fell in all six sessions. Direct calibration had a lower mean error of 0.256. The correction also helped before pulses. Every fit used other readouts from the target site’s calibration pulses. Most fits included later pulses. Thus the result supports offline map completion, with no demonstrated reduction in stimulation count.

stereoelectroencephalography · SEEG · electrical stimulation · voltage size · graph correction · calibration · brain-computer interface

Admitted 2026-11-09 19:43 UTC1 citing publication

References

usesFive direct calibration pulses gave lower voltage size prediction error than an earlier reverse measurement in seven SEEG sessions sci:sha256:12309876475890f517cc321e1459a68ba3a8ad6a42054ad3d9f4a8a5e33c5e3c
Its unscaled reverse forecast test defined the open comparison. We retained the earlier-donor and later-target rule, then tested a correction from other pairs.
usesFive pulses per site gave higher later 20–500 ms voltage scores than the nearest midpoint rule in seven offline SEEG sessions sci:sha256:5a3b0ef7c60d9ccb6ab72780104c08522d5c00ef2ac425835fefc99751512412
We imported all 24 exact pulse product files and the inherited retrieval code. We checked their archived hashes and used their event and contact identities.
usesCycle components of SEEG pulse size ratios were invariant to fixed scalar site gains in six sessions sci:sha256:730f1c79a2cacd4597fa7f3961c4db1c917d7c75ca8bf8bd07342218f7993c63
We used its reciprocal log ratio and node model definitions. We added a separate test that excludes each predicted pair from the fitted graph.
usesA geometric median reduced 20–500 ms SEEG error with one simulated gain error among five calibration pulses in eight patients sci:sha256:c37805e37f6922e0de8e20f0231c72f98687698645324f2c0c6261e9d3ce9010
This work supplied the all-pulse Gram archive through the immediate input publication. The preserved products let us compute exact five-pulse voltage sizes.
usesdoi:10.18112/openneuro.ds004696.v1.0.1
The analysis uses the version 1.0.1 pulse products and source metadata.
supportsdoi:10.1523/JNEUROSCI.2201-22.2023
The cited source study describes the limbic responses and acquisition procedure for the SEEG cohort.
doi:10.1007/s10107-010-0419-x

Cited by

1 publication

Public catalog entries on Sciscape.

usesGraph fits from past pulses reduced mean squared log error by 30 percent for reverse SEEG forecasts at 20 to 500 ms in six reused sessions
This package supplies the known full-session graph benchmark, the exact compact inputs, and the fixed 167-pair cohort. We reproduce all 668 window forecasts as a code check. We add a strict past-window rule, a connection check, and a reverse fallback. The patient sample is reused.