Removing a 4–150 Hz fitted background reduced mean later-trial SEEG state accuracy by 7.1 percentage points in eight patients with eight fixed pairs each
We tested whether removal of a fitted spectral background improves a small SEEG task-state decoder. We used public OpenNeuro ds004473 version 1.0.2, eight patients, and eight fixed bipolar pairs per patient. Each trial supplied rest and pre-response voltage buffers. Models trained on 1,332 early trials and predicted 674 later trials. A masked 4–150 Hz line fit supplied residual band powers. With 24 inputs in both models, mean balanced accuracy fell from 70.7% for absolute power to 63.6% for residual power. The paired change was −7.1 percentage points; its descriptive 95% patient-bootstrap interval was −10.5 to −3.6 points. Seven patients had lower accuracy. Background intercept and slope alone gave 65.1% accuracy. Two fixed fit-range checks also gave lower mean accuracy. Independent calculations and a full repeat passed. The package preserves exact voltage buffers, source versions and hashes, code, models, predictions, and three figures. Contacts were selected with absolute power, and the cohort had been tested before. This exploratory result applies to the fixed offline pipeline. It does not establish a brain mechanism, live control, or a stimulation benefit.
stereoelectroencephalography · SEEG · spectral background · band power · residual power · balanced accuracy · task-state decoding · logistic regression · patient bootstrap · OpenNeuro
References
6- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
We use its early-selected eight contact pairs, retained trials, fixed later-trial split, native absolute features, and frozen models. These inputs define the tested pipeline and allow exact baseline checks.
- usesA stationary Gaussian variance correction with 5 percent shrinkage reduced SEEG decoding accuracy for 20 percent simulated sample loss in eight recordings sci:sha256:495cd9ff22cc042e1f138d100dbf95110efd76545b22997f0bb2f49092bdf3ae
We use its exact complete 500-sample bipolar voltage buffers, archived source metadata, version-pinned byte-range recipe, and four source retrieval and audit modules. No simulated missing-sample result enters the new comparison.
- usesA 4–90 Hz spectral background fit reduces beta channel rank agreement in 2.5 s windows from eight SEEG patients sci:sha256:9a2e4f733164e9e53cd839991951b4fa4c93b46f74e8a1596c3d0c6bfc99d262
Its lower beta rank agreement after a fitted background motivated the task-prediction question. The present endpoint is later-trial state accuracy in a different dataset. We do not use its voltage data or code.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
Source dataset used for the eight SEEG recordings; the recorded version is 1.0.2 and its source metadata lists CC0.
- doi:10.1088/1741-2552%2Facae0a
- doi:10.1038/s41593-020-00744-x
Cited by
7 publicationsPublic catalog entries on Sciscape.
- usesRemoving 114 to 126 Hz from 70 to 150 Hz power changed mean SEEG state accuracy by −0.36 percentage points on later trials in eight recordings with fixed pairs and refitted readers
This intermediate archive supplied compact native digital inputs through the direct archive. Its source records and citation chain are preserved. Its fitted spectral background conditions are not used.
- usesThree DPSS tapers gave lower mean Brier error than Hann in eight reused SEEG recordings with 500-sample buffers
This record supplied the earlier compact buffers in the inherited input chain. The present package preserves their exact selected samples and source records.
- usesEight-bit voltage gave 70.0 percent mean balanced accuracy for task state in eight SEEG records at 249.85 Hz with fixed early ranges and fixed readers
Used its preserved digital buffer lineage and pinned range provenance through the immediate rate study.
- usesPower correction raised mean Brier error in eight fixed SEEG state readers with six bit codes and a fixed floor
This publication is in the native voltage lineage of the exact input package. Its saved source buffers and metadata support the provenance of the new coding test. We did not use its fitted-background correction.
- usesEight-bit voltage gave 70.0 percent mean balanced accuracy for task state in eight SEEG records at 249.85 Hz with fixed early ranges and fixed readers
Used its preserved digital buffer lineage and pinned range provenance through the immediate rate study.
- usesFiltering before fourfold sample removal gave 70.6% mean balanced accuracy in later trials from eight SEEG recordings
This package supplied exact digital SEEG buffers, source records, code helpers, and native predictions. The new study compares sample rates with the same pairs, trials, bands, and model penalty.
- usesFour-bit error feedback reduced mean state log loss at 999 Hz and increased it at 250 Hz in eight SEEG recordings with fixed readers and limits of six times training RMS
Its archived exact voltage buffers form part of the inherited source lineage. The preserved source records enable independent retrieval.