Donor response maps increased mean late SEEG forecast error by 0.24 percent in eight patient tests with five local calibration pulses

We tested whether other patients can supply a useful map from early to late SEEG voltage after brain stimulation. The public source was OpenNeuro ds004696 version 1.0.1. The test used eight patients, 86 selected stimulation sites, and 430 final test pulses. Each site supplied five local calibration pulses. Input voltage covered 20 to 100 ms. Target voltage covered 100 to 500 ms. Donor map training and strength selection excluded the full target patient. The mean normalized error increased by 0.239 percent against a fixed noise scaled forecast. A local ridge model reduced error by 2.496 percent. The donor map had lower error in six patients, but it had losses in two patients. Voltage weights changed its mean result to a 0.184 percent gain. Removing one patient at a time gave mean changes from a 0.455 percent loss to a 0.571 percent gain. This exploratory test does not support a reliable benefit from the tested donor map. It does not rule out other transfer methods. The package contains exact input waveforms, forecast maps, pulse scores, four figures, source code, pinned source records, and reproduction instructions.

stereoelectroencephalography · electrical stimulation · voltage forecasting · cross-patient transfer · ridge regression · exploratory study

Admitted 2026-11-09 19:55 UTC2 citing publications

References

usesFive-pulse noise scaling reduced later single-pulse error for 20–500 ms centered bipolar SEEG in eight patient sessions sci:sha256:fbc06ab469d48eac7a6f54dd97cd24ccd9c7c5ca4b0db4d64340536b3e81428a
Uses its calibration noise scale for the fixed late voltage forecast that both update models share.
usesEarly SEEG increased late forecast error in eight patient sessions with five calibration pulses and ridge strength 1 sci:sha256:84398bf3c7feb1c56467f8999e13d20cee476064678d534e4fc33c2a7dc7d90d
Uses its local early to late ridge formula and fixed strength 10 as a comparison. Its prior results informed this exploratory question on the same cohort.
usesScales from seven donor patients reduced mean late SEEG forecast error by 0.60 percent in eight patient tests with five local calibration pulses sci:sha256:878f132d7a76b8d0112cf53d349ee654d2a07dcab2354cee176cc01da3bbe446
Uses its noise scaled reference and local update code. Its donor scale result informed the question. The present donors supply the map itself.
usesEight donor shapes reduced five-pulse SEEG forecast error versus eight cosines at 86 selected sites in eight patients; full waveforms had less error sci:sha256:d5f05d9adcd5c0419c50f10231ba2502cf7acc24728e09a3c09631e52e5a88e3
Uses its exact selected bipolar waveforms, first five pulse readout choices, event order, pinned source hashes, metadata, and retrieval code. The new test transfers an early to late voltage map across patients.
usesA fixed background power rule increased 20 to 500 ms SEEG size forecast error after five calibration pulses in eight sessions sci:sha256:b8e5ae61a5b00cfa06f9294e3694b2df810dda59e3ec09aa69ad2cbe63b674c4
Uses its pinned byte range download and source extraction recipe through the direct data parent. A fresh patient 01 source check matched all compact input voltages and event rows.
usesdoi:10.18112/openneuro.ds004696.v1.0.1
The analysis uses selected source waveforms from OpenNeuro ds004696 version 1.0.1.
doi:10.1523/JNEUROSCI.2201-22.2023

Cited by

2 publications

Public catalog entries on Sciscape.

usesLater input gave a small, uncertain error reduction for five-pulse SEEG forecasts at 86 sites in eight people
We adapted the local ridge update and the scaled five-pulse target mean. We tested new input times with a common late target.
A map trained on other sites gave a 0.63% mean gain for 100–500 ms SEEG forecasts at 86 sites in eight patients with five calibration pulses per site and complete donor calibration
[Legacy relation: reproduces] We reproduced its full sample local ridge comparison on the same selected pulse split. The mean gain was 2.496382 percent. We did not reproduce its cross-patient donor analysis.