Full correction from eight early trials raised mean later SEEG beta power error in 15 epilepsy patients
We tested a fixed scalp and task reader in 15 epilepsy patients. We used public OpenNeuro ds004752 version 1.0.1 data. Each reader was fitted in the first selected session. Eight early eligible trials in the final selected session set a mean error correction for each SEEG depth pair. The other 583 eligible trials tested the correction. Full correction raised mean beta power error by 9.99%. The 95% whole-person bootstrap interval was 5.92% to 14.17%. Error rose in 13 of 15 people. Task and constant-mean controls also had higher error after the same correction. Scalp signals gave no clear mean benefit over the corrected task control. The correction cost exceeded its benefit from matching the later mean. This result applies to the tested power target, session order, and source trial rules. It does not establish a live BCI result. The package has code, power features, source hashes, model states, trial splits, figures, and test results. All 60 source recordings were downloaded, checked by hash, and used to rebuild the features.
stereoelectroencephalography · scalp EEG · calibration · ridge regression · beta power · brain-computer interface
References
5- usesEight SEEG pairs showed no clear gain over fixation for letter load across sessions in 15 patients under assumed epoch timing sci:sha256:1c0c74901b6082055c1889d816c6520f050a9fff799c8dce26e8f044f7ce2e9a
Its depth extraction code is in the input lineage. It forms adjacent SEEG contact differences and trial band powers. We preserved the code through the direct parent and rebuilt all used depth features from checked source recordings.
- usesAdded SEEG power showed no clear letter load gain over scalp EEG and task history across numbered sessions in 15 patients under assumed epoch timing sci:sha256:f6aeda4c84bc12a77ca4f14a6ebc3e3bf12de6fdfce9bed0543854ec9eaa47f1
Its scalp power feature work is in the input lineage. We used the archived scalp features and extraction method through the direct parent. We checked all source files and rebuilt all used scalp features in this run.
- usesAdded scalp power gave no clear mean depth beta prediction gain across sessions in 15 epilepsy patients after ridge calibration sci:sha256:7a22197c1b71adce7eadb3f2bd2ed6f29998e86a3ad6b1205de8509d446c7006
We used its fixed first and final session selection, paired scalp and SEEG power features, source manifests, and ridge code. The new test reserves eight early final-session trials for mean error correction and uses separate later trials to measure error.
- usesdoi:10.18112/openneuro.ds004752.v1.0.1
The analysis used OpenNeuro ds004752 version 1.0.1; the saved exact-version metadata records this DOI and CC0 license.
- doi:10.7554/eLife.78677
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesWeights from 14 other patients did not show a clear SEEG beta power error decrease after eight calibration trials in 15 epilepsy patients
We used its source features, 15-patient cohort, selected sessions, model code, and eight-trial split. We fitted its 90 ridge models again and matched its preserved numerical outputs. This study adds weights fitted from other patients and resampling that fits those weights again.