Three DPSS tapers gave lower mean Brier error than Hann in eight reused SEEG recordings with 500-sample buffers

We tested whether three spectral tapers improved a reader of two task states. The public source was OpenNeuro ds004473 version 1.0.2. We reused selected contact pairs and exact voltage buffers from eight people. Each 500-sample buffer lasted about 0.5 seconds. Four power estimates used the same buffers, bands, trial split, and logistic model rule. Each reader learned from early trials and predicted later trials. Three DPSS tapers gave mean test Brier error 0.177218. Hann gave 0.190139. The paired change was -0.012921, with a 95% person bootstrap interval from -0.021412 to -0.004048. Seven of eight recordings had lower error. The test used 674 later trials. Each person had equal weight in the means. Gaussian controls showed lower power variation with three tapers. Other controls showed changed time weights and frequency responses. These controls do not identify the cause of the decoding gain. This exploratory study reused earlier test trials. Its labels describe event-aligned task states under offline eligibility rules. It does not establish live control or stimulation benefit. The package preserves exact inputs, code, source versions and hashes, features, models, predictions, uncertainty draws, three figures, and reproduction instructions. Independent checks passed. A full replay matched all 31 numerical outputs exactly. Fresh source checks matched trial ranges for all eight people and all retained input arrays for one recording.

SEEG · spectral estimation · DPSS · Hann taper · Brier error · brain-computer interface

Admitted 2026-11-09 19:58 UTC12 citing publications

References

usesFiltering before fourfold sample removal gave 70.6% mean balanced accuracy in later trials from eight SEEG recordings sci:sha256:8bc11c5e9496aba0a89fbf8535edb5ce1336d5a00cc9df2b9795ddfb7b8ddde4
This package supplied the exact digital buffers, fixed selection, source metadata, numerical helpers, and retrieval code. We used them for a matched spectral comparison.
usesRemoving 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 sci:sha256:c3469f7de479fc287d2128fe8ba0836892261cc3bf0dc5c837eb721c6ef303b1
This record supplied the earlier compact buffers in the inherited input chain. The present package preserves their exact selected samples and source records.
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 record fixed the original trial eligibility and early contact pair selection. Its EDF helper code supports the inherited retrieval method.
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
This record supplied the earlier digital voltage extraction and pinned source range records. Those records support fresh source checks and input retrieval.
usesdoi:10.18112/openneuro.ds004473.v1.0.2
This versioned public dataset supplied the SEEG recordings and source metadata used for the eight-person matched comparison.
doi:10.1088/1741-2552%2Facae0a

Cited by

12 publications

Public catalog entries on Sciscape.

usesGamma envelope concentration lowered mean later-trial Brier error against three band powers in eight reused SEEG recordings with 500-sample buffers
We reused its exact 500-sample bipolar buffers, fixed selection files, source headers and range records, and numerical helper code. Its three-taper power comparison informed the stronger baseline.
Recent sample weights gave lower mean Brier error than reversed weights in eight reused SEEG recordings with 500-sample buffers
[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.
usesRecent sample weights gave lower mean Brier error than reversed weights in eight reused SEEG recordings with 500-sample buffers
Supplies exact digital buffers, fixed selected pairs, source metadata, retrieval helpers, and numerical fitting code. The new study tests mirrored time emphasis.
usesAdded imaginary coherency increased mean Brier error for tuned linear readers in eight SEEG records with fixed 500-sample inputs
This publication supplied the exact short waveform and pinned raw byte-range retrieval lineage inherited through the immediate input publication.
usesA Hann gamma envelope feature did not lower mean Brier error beyond matched power on later trials in eight reused SEEG records
This work supplied the inherited DPSS power code and the versioned EDF range retrieval method. We used three DPSS tapers in the common power baseline and kept the exact selected-waveform lineage.
usesSigned gamma timing gave an uncertain Brier error change in eight reused SEEG recordings with 500-sample buffers
This study supplied the upstream 500-sample waveform package and the three-DPSS power comparison. The new study uses a symmetric three-taper power control and fixed penalties.
usesA Hann gamma envelope feature did not lower mean Brier error beyond matched power on later trials in eight reused SEEG records
This work supplied the inherited DPSS power code and the versioned EDF range retrieval method. We used three DPSS tapers in the common power baseline and kept the exact selected-waveform lineage.
usesThree tapers increased mean SEEG gate score at error cost 4 in eight reused recordings, but the change interval included zero
This package supplied exact voltage buffers, fixed contact pairs, source records, and reader rules. We reproduced its two readers and tested fixed-cost gates.
usesEight offline SEEG readers with median gamma power had 0.500 mean worst contact Brier error under simulated 1 mV, 100 Hz faults of 250 samples
This publication supplies the exact digital voltage buffers, pinned source records, and MIT numerical and retrieval code.
usesOne random taper mixture had higher mean Brier error than a matched taper at C=1 in eight reused SEEG records
This publication supplied the DPSS settings and the matched single-taper control on the same test periods. We reproduced its three baseline Brier errors and added random taper rules at one- and two-transform budgets.
usesOne shared gain constraint gave lower mean Brier error than eight pair gain constraints in eight reused SEEG records under a uniform gain model
This study supplied the exact short-buffer source lineage and the pinned EDF range retrieval recipe used to rebuild every selected trial.
usesTraining for simulated fixed 0.8–1.25 pair gains reduced worst loss and mean clean accuracy in eight SEEG readers
We reused its exact eight digital SEEG buffer files, selection files, source records, symmetric Hann feature rule, and range retrieval helpers. The new work tests simulated pair gains and gain training on those fixed inputs.
usesThe last 250 samples gave mixed Brier error changes against 500 samples in eight SEEG task state readers with three tapers
This study supplied the exact 500-sample digital buffers, source records, spectral helper, model helper, and retrieval code. We kept its full-window rules and tested first and last halves.