Adding spectral slope increased mean later Brier error in eight SEEG recordings with fixed pairs and penalty 1
We tested spectral slope in eight SEEG recordings from OpenNeuro ds004473 version 1.0.2. The study kept eight bipolar pairs per person. Early trials set the logistic models. Later trials gave 1,348 test windows. The primary model used three DPSS tapers and penalty 1. Adding eight slopes to 24 power inputs raised mean Brier error from 0.17597 to 0.17914. The paired mean change was +0.00317. Its descriptive 95% person bootstrap interval was +0.00037 to +0.00605. The exact sign-flip reference was 0.09375. Mean accuracy rose from 73.51% to 74.16%. Added slopes raised mean Brier error in all six fixed spectrum and penalty settings. A matched ratio control gave a similar result. A proof and checks show that constant positive pair gain preserves shape inputs. All slope-only decisions stayed fixed. The package preserves source buffers, code, fits, probabilities, figures, and rerun commands. This exploratory test uses two task states and reused later trials. It tests no live commands, new session, or stimulation.
spectral slope · SEEG · stereoelectroencephalography · Brier error · brain-computer interface · probability prediction · DPSS · power features
References
6- 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 sci:sha256:2b6137f1e4b7d2447d151534f11a2b3b8f441e2892c12dbf1dfe3e47f954dd1a
This archive supplied the exact digital buffers, fixed selection files, source metadata, and retrieval code. The new study computes slope and ratio features and fits new readers on those inputs.
- usesFiltering before fourfold sample removal gave 70.6% mean balanced accuracy in later trials from eight SEEG recordings sci:sha256:8bc11c5e9496aba0a89fbf8535edb5ce1336d5a00cc9df2b9795ddfb7b8ddde4
Its source-rate voltage buffers and range helpers are preserved through the immediate input archive. The present study uses 500-sample buffers without reduced-rate conditions.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
This study keeps its early-trial choice of eight bipolar pairs and its retained trial order. The result is conditional on that power-based pair choice and retrospective quality selection.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis uses the source-rate voltage buffers and task records from OpenNeuro ds004473 version 1.0.2.
- doi:10.1088/1741-2552%2Facae0a
- doi:10.1038/s41593-020-00744-x
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesAdding a band contrast reduced mean test error beyond features outside the band in eight SEEG records
Used its in-band slope comparison to define a different question. The new reference reader uses frequencies outside 70 to 110 Hz. The new test asks whether a band contrast adds predictive value after that reference.