sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

Low correlation selection increased later feature dimensions in eight SEEG records. Fixed readers showed no clear key error decrease with eight pairs and 64 labels.

This study tested SEEG pair selection for low feature correlation. Selection used no key labels. It used public OpenNeuro ds004473 v1.0.2 and all eight people. Each signal reader used eight bipolar pairs and three power bands. All rules used the same earlier label samples and fixed logistic readers. The main test used 64 labels, C = 0.01 and 50 seeds per person. The low correlation rule increased the mean later feature dimension score from 13.92 to 17.88. Its mean Brier error difference from random selection was -0.00107. The 95% person bootstrap interval was -0.00338 to +0.00123. The exact sign-flip value was p = 0.4141. Thus, the test gave no clear key error decrease. It did not show equal performance. A constant reader also had similar mean error. The package preserves source versions, feature data, code, all reader states, results and figures. Fresh raw checks and an exact analysis repeat passed. This is an exploratory offline result in a reused clinical cohort. It does not establish continuous BCI performance or stimulation effects.

SEEG · brain-computer interface · feature selection · correlation · Brier error · intracranial EEG

Admitted 1 citing publication

References

6
  1. usesSelection for key identity had lower mean later key error than activity selection in eight offline SEEG records with tuned readers sci:sha256:8f807b882f06ccced0e3c3ddf31d99bf21b9b7ca83164353fda7c8a2ab07e7f8

    This study used the recorded key mapping and class-contrast pair score from that work. Its earlier task comparison helped define the matched key-score reference. The present main rule uses no key labels for pair selection.

  2. usesSelection by correlation gave more normalized gamma voltage dimensions than selection by power for four SEEG channels in 33 patients sci:sha256:deaea2ba2cbb661f8ed6a6384f88b4aebfc8a1760a9c8c41c0253f74c479f39b

    Its correlation selection result in resting SEEG motivated a task test of dimension scores. We used none of its data. We tested power features, eight pairs and later key error, which differ from its voltage features and four-channel test.

  3. 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 publication supplied the original five-window feature source and extraction code. We preserved its source metadata, pinned raw range recipes and code. We rebuilt all 202 raw trials for one person and checked sampled trials for all eight people.

  4. usesUncertainty sampling had higher mean later key probability error than random sampling at 64 labels with fixed linear readers in eight offline SEEG records sci:sha256:9d4bd48a6082903c9011857e0b266ed71af75d1fdb69a3d187bd2b8de5cca23f

    This study reused the exact before-response feature projection, trial records, split and raw audit inputs. Its prior reader results informed the fixed C choice. The new analysis compares pair selection rules with common labels and pool scaling.

  5. usesdoi:10.18112/openneuro.ds004473.v1.0.2

    Analyzed source dataset: OpenNeuro ds004473 version 1.0.2.

  6. doi:10.1088/1741-2552%2Facae0a

Cited by

1 publication

Public catalog entries on Sciscape.