Recent sample weights gave lower mean Brier error than reversed weights in eight reused SEEG recordings with 500-sample buffers
We tested whether recent sample weights improve offline task-state reading against their earlier-time mirror. The data were 674 later trials from eight SEEG recordings in OpenNeuro ds004473 version 1.0.2. Both estimates used the same 500 samples, contact pairs, bands, and reader rule. Their taper frequency magnitudes were equal. Mean Brier error was 0.170244 for recent weights and 0.204975 for the mirror. The equal-person paired change was -0.034732. Its descriptive 95% person-bootstrap interval was -0.060127 to -0.011236. Seven people had lower error. Against centered weights, the change was -0.007027, with an interval of -0.022926 to +0.008271. That secondary result does not establish an improvement. This is an exploratory test in a reused clinical cohort with known event times and offline eligibility. It does not establish live control, shorter delay, a neural source, or a stimulation effect. The package preserves exact inputs, source pins, code, all 40 readers, predictions, three figures, and reproduction instructions. Independent implementations matched the decisions. All 34 replayed outputs matched exactly. Fresh source ranges, a complete retained recording, and their checks are preserved.
SEEG · brain-computer interfaces · spectral estimation · time weighting · Brier score · task-state decoding · DPSS · reproducibility
References
6- usesFiltering before fourfold sample removal gave 70.6% mean balanced accuracy in later trials from eight SEEG recordings sci:sha256:8bc11c5e9496aba0a89fbf8535edb5ce1336d5a00cc9df2b9795ddfb7b8ddde4
Provides the earlier exact-buffer and pinned public-source retrieval chain used by the immediate parent. The new study keeps the original 500-sample inputs.
- Three DPSS tapers gave lower mean Brier error than Hann in eight reused SEEG recordings with 500-sample buffers sci:sha256:6b6f409ccedb538d81478030fd7ac7ab8512d1e3bcef6a9af4feebff6c7e2f5e
[Legacy relation: reproduces] The Hann and periodic three-DPSS references reproduce all eight saved test-logit arrays exactly. This checks two parent estimates, not an independent cohort.
- usesThree DPSS tapers gave lower mean Brier error than Hann in eight reused SEEG recordings with 500-sample buffers sci:sha256:6b6f409ccedb538d81478030fd7ac7ab8512d1e3bcef6a9af4feebff6c7e2f5e
Supplies exact digital buffers, fixed selected pairs, source metadata, retrieval helpers, and numerical fitting code. The new study tests mirrored time emphasis.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
Provides the inherited eight bipolar pairs per person and offline trial eligibility rules. The new study keeps both choices fixed.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis uses the public OpenNeuro ds004473 version 1.0.2 dataset; saved source metadata and the version-pinned retrieval records identify this exact dataset release.
- doi:10.1088/1741-2552%2Facae0a
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesSigned gamma timing gave an uncertain Brier error change in eight reused SEEG recordings with 500-sample buffers
Its recent-versus-reversed sample weight test informed the question about timing direction. The new study directly tests a signed timing feature against matched unsigned timing controls. It does not reuse its scores as new evidence.
- usesTwo 250-sample spectra gave lower mean Brier error than one 500-sample spectrum in eight reused SEEG records with equal sample weights
Its prior sample-weight test motivated an equal-weight primary comparison and a separate Hann weight control.