Fixed early SEEG readers with eight inputs had 40% mean balanced accuracy for three force target ranks across executed grasps in four people

We tested whether fixed SEEG readers transfer three force target ranks between executed pinch and power grasps. The public data contain 32 blocks and 472 retained trials from four people with epilepsy. Blocks 1 and 2 trained each reader. All later blocks were test blocks. The primary reader used eight inputs and four frequency bands at 0.5 to 1.5 seconds after the target change. Its mean balanced accuracy was 39.9% across grasps and 34.3% within grasp. Its mean balanced log loss was worse than uniform predictions. The task-versus-rest control reached 80.2% across grasps. An added late sensor control reached 83.3% within grasp. Task-state accuracy alone does not establish three-target readout quality. This small study tests one fixed method. It does not establish a population benefit from transfer or rule out other SEEG methods. The package preserves code, features, three figures, all test predictions, source records, and repeat instructions. All 1,888 used raw windows passed a new source check. A complete analysis repeat matched all saved analysis outputs exactly.

stereoelectroencephalography · brain-computer interface · grasp · force target · cross-task transfer · balanced accuracy · logistic regression

Admitted 2026-11-09 18:21 UTC1 citing publication

References

usesFixed SEEG readers scored 80% mean balanced accuracy early and 54% later across executed grasps in four people sci:sha256:9317803e3e59d0db5ce4f76b02220e1344da603fb253dca80e67acfbce823766
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 sci:sha256:36e4b7ae7abe3acc0dcb12020c08825c3fb662eb2187c32e1470bc56dd013466
Uses the force-sensor baseline finding as context for the planned sensor control. This publication does not supply the new fitted models or scores.
usesRest calibration reduces mean task accuracy in fixed SEEG imagery tests from two people sci:sha256:b2d5a1ce684c63615eb48aace1fd2a51169cfeb20f8e08e74a6759aa6964ec06
Uses the A/B feature extraction and source records inherited through publication 2526d00e-a871-4f0c-b786-e96c8ccd8647. The package preserves the original license notice. This work enabled the four-person feature set used for the new target test.
usesPaired rest power did not show a clear three-target SEEG gain from actual to imagined grasp with eight inputs in two people sci:sha256:a0079e444a347fd44ee544b204313f3368e92b229a4b16627769398c0ca00dac
Uses the earlier two-person executed-to-imagined three-target result as context to choose a distinct test: transfer between two executed grasps in four people. The earlier scores are not new observations in this study.
doi:10.1371/journal.pone.0150359
usesdoi:10.5061/dryad.nq4fs
The NEMAR release analyzed here is a conversion of the public Dryad SEEG data.
usesdoi:10.82901/nemar.nm000397
Analyzed source is NEMAR nm000397, version 1.0.0, as recorded in the source manifest and source links.

Cited by

1 publication

Public catalog entries on Sciscape.

usesEarly linear SEEG corrections increased mean relative squared error by 66.5% over cue and early sensor forecasts during grasps in four people
This study reused its 32 feature blocks, eight raw audit trials, source metadata, retrieval receipts, and extraction code. These files supplied all 472 eligible grasp events.