Donor slopes raised mean SEEG phase error in late delay after 16 early delay trials per session in 13 patients under assumed trial order

We tested a fixed method to reduce SEEG decoder calibration. The data were OpenNeuro ds006136 version 1.0.1. The test had 13 patients, 14 sessions, and 751 later trials. A donor model used only the other 12 patients. Each local model used 16 paired fixation and early delay trials per session, plus an initial fixation reference. The test compared fixation with late delay. The donor slope reference raised mean Brier error from 0.258567 to 0.259975. Only 2 of 13 patients had lower error. The paired error improvement was -0.001408. Its descriptive 95 percent patient bootstrap interval was -0.002682 to -0.000321. Donor training folds overlap. This interval does not give an independent population test. Both models had more mean error than a constant 0.5 prediction. The result applies to the fixed features, penalty, calibration trials, and assumed source trial order. The package contains three figures, all planned conditions, source hashes, features, code, predictions, and reproduction instructions. All 14 EDF files and 103 metadata files passed source hash checks. Source re-extraction and a full analysis rerun matched the saved files. The study tests task phase readout. It does not test memory content or brain stimulation.

SEEG · brain-computer interface · calibration · transfer learning · Brier error · working memory · phase decoding

Admitted 2026-11-09 17:04 UTC1 citing publication

References

usesAn SEEG model trained on fixation and early delay gave 51.9 percent mean balanced accuracy on fixation and late delay in 13 held-out patients with baseline calibration under source trial order sci:sha256:c925d9ba47fc49fc734b45e9ba8433a2ca6f2950272637d7ff9b9207c5e3d8c9
This publication supplied matched regional power tables and the fixed baseline scale, donor weights, and source audit helpers. We reproduced its three donor model score sets and added fixed local calibration tests.
usesEarly delay SEEG power gave no mean Brier gain in later recall forecasts for 13 patients under source trial order sci:sha256:adbcbe06fcce87132c9633b5137a24af646e5c49afc3ac2d09469a1e05618474
This publication supplied the unchanged EDF retrieval and extraction code. We used that code to retrieve and hash all 14 source EDF files and rebuild the matched features.
usesdoi:10.18112/openneuro.ds006136.v1.0.1
The study used OpenNeuro ds006136 version 1.0.1, with source files checked against pinned hashes.
doi:10.1016/j.neuroimage.2026.121718

Cited by

1 publication

Public catalog entries on Sciscape.

usesDonor slopes raised mean SEEG phase error in late delay after 16 early delay trials per session in 13 patients under assumed trial order
This publication has the same study data, models, scores, and figures. This version corrects the package check to exclude the retrieval metadata directory added by sci fetch. All study files remain subject to hash checks.