Fixed SEEG readers scored 80% mean balanced accuracy early and 54% later across executed grasps in four people
A fixed reader used public SEEG inputs to separate rest from executed hand grasp periods. The test used four people, two grasp types, and 236 trials in 16 separate test blocks. The first two blocks of each grasp trained each reader. Both readers then scored the same target trials. Mean balanced accuracy was 79.18% with training on the same grasp and 80.21% with training on the other grasp. The mean difference was +1.03 percentage points. The descriptive four-person bootstrap interval was [-2.72, +7.08] points. A planned timing check gave 54.00% accuracy in the later task window with the same readers. Four people cannot establish population transfer or equivalent performance. The package provides source receipts, features, raw audit trials, fixed models, predictions, code and figures. The inputs lack contact maps. The test concerns task periods with visible cues and movement feedback. It does not establish spontaneous intent, live control or a brain stimulation method.
stereo-EEG · SEEG · brain-computer interface · grasp transfer · balanced accuracy · neural decoding · epilepsy
References
4- usesRest calibration reduces mean task accuracy in fixed SEEG imagery tests from two people sci:sha256:b2d5a1ce684c63615eb48aace1fd2a51169cfeb20f8e08e74a6759aa6964ec06
Reuses 16 execution feature files, four raw audit files, metadata and source receipts for A and B. Adapts its finite-window extraction method for C and D. The new study tests transfer between executed grasps in all four people.
- usesdoi:10.82901/nemar.nm000397
The analyzed recordings are NEMAR nm000397, version 1.0.0; the publication records the manifest URL and hash.
- usesdoi:10.5061/dryad.nq4fs
The NEMAR conversion used for this analysis derives from the Dryad SEEG dataset; the original release is identified in the source records.
- doi:10.1371/journal.pone.0150359
Cited by
7 publicationsPublic catalog entries on Sciscape.
- usesFixed early SEEG readers with eight inputs had 40% mean balanced accuracy for three force target ranks across executed grasps in four people
Reuses 32 executed-grasp feature blocks, eight raw audit trials, source metadata, range receipts, and extraction code. These files support the new three-target test across executed grasps. The new task-versus-rest control also repeats the parent method.
- usesSEEG readers with eight inputs increased squared sensor error over target estimates during late grasps in four people
Reuses all 32 executed task feature blocks and eight raw audit trials. Also reuses source metadata, receipts and extraction code. These inputs support new sensor estimates in separate blocks. The displayed target supplies the reference estimate.
- usesEarly linear SEEG corrections increased mean relative squared error by 66.5% over cue and early sensor forecasts during grasps in four people
The direct feature parent used its extraction code, audit windows, and source records. This package preserves and adapts that extraction code and keeps its license notice.
- usesMixed training reduced the larger early or late test Brier error from 0.325 to 0.255 with equal window counts in SEEG data from four people
We inherited the four-person executed-grasp features, exclusion rules, and training and test block split through the direct parent. Fresh extraction from all 32 complete public signal records matched the compact feature arrays exactly.
- usesTraining on late grasp windows raised mean late test balanced accuracy from 53% to 64% with eight SEEG inputs in four people
This study supplied the exact compact features, raw audit files, source metadata, extraction code, and block split. Its early and late test scores led to this question. We fit late task readers with the same inputs and score the same test blocks. The early balanced accuracy and all eight input sets match the parent.
- usesSEEG readers with eight inputs had greater mean target loss than uniform predictions during late grasps in four people
This study uses the whole-block comparison of same-grasp and other-grasp training. It extends binary task detection to three active target levels in the late task window.
- usesTraining at two grasp times reduced offline SEEG probability error in four people with eight inputs and C = 0.1
Reused all 32 executed-grasp feature records, eight raw signal audit files, source metadata, pinned manifest, retrieval receipts, and the extraction code. These artifacts enabled the same block split and an Early baseline. This study added weighted training at two task times, fixed-input and equal-row controls, model-setting checks, and fresh-source replication.