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

Fixed models with past SEEG power gave no clear Brier score gain over cue inputs for 100 ms keypress forecasts in eight later task segments

We tested whether past SEEG power adds value beyond known cue inputs for a recorded keypress forecast. The target was a keypress in the next 100 ms. The source was OpenNeuro ds004473 version 1.0.2. Each of eight patients had separate fixed logistic models. Early data fit the models. Middle data chose a penalty. Later data supplied 339 test trials and 2,371 time points. Each trial and patient had equal weight. Mean Brier score was 0.1280 with cue inputs and 0.1300 after adding 32 SEEG power features. Lower is better. The paired difference was +0.00207, with descriptive 95 percent patient interval [-0.00627, +0.01202]. Five patients improved. This gave no clear primary gain. The cue-only mean ROC AUC was 0.868. The 250 and 500 ms comparisons also gave no clear Brier gain. Secondary log loss and AUC did not all favor the same model. These reused data test later segments in the same recordings. Three patients changed cue condition. The result does not establish equal model performance or no useful neural signal. The package preserves three figures, powers, source joins, code, all forecasts, and pinned raw-data recipes. Fresh EDF checks, independent solvers, and an exact complete repeat passed.

stereoelectroencephalography · keypress forecasting · Brier score · brain-computer interfaces · cue inputs · neural power

Admitted 2 citing publications

References

7
  1. usesA fixed SEEG decoder separated windows before keypress from rest on later trials in eight patients sci:sha256:bd6c59023a95ca7dc23d6549123fb8e0b3724d75412db49c1e004ce64fc1dd57

    This work supplied the earlier response-versus-rest selection of eight bipolar pairs per patient and the four power bands. We kept the pairs fixed and checked that their selection data preceded the scans.

  2. usesFixed SEEG commands showed no clear 500 ms timing gain over local delay swaps in eight recordings sci:sha256:b5617cb10086e0cf1627c0451270d72ed632b277ca3fc20c888444f86f77c340

    Its local delay-swap result motivated an explicit cue-clock forecast baseline. The present test fits new probability models and compares later Brier scores, rather than repeating its command permutation test.

  3. usesAt a zero threshold, eight-pair SEEG decoders gave a mean of 36.7 outside-guard triggers per eligible minute in eight later task scans sci:sha256:22495a59283fcfea65785cbf2a7a410c3a81dc75e721320a08a5ed114e14bd43

    This work created the continuous past-window power archive later preserved in the direct source package. We used its sample endpoints, inherited event retention, and raw feature recipes.

  4. usesSEEG key decoding from the first fast block had a mean AUC of 0.54 in later slow blocks in six patients sci:sha256:6407f4f45a2db401fef737a5f946c2656d5af25ba66dbcab30e4417db2c4fc72

    Its report of a source key-timing concern for patient 11 motivated the fixed sensitivity test that omits this patient.

  5. usesMoving-minute calibration left more than three background commands per eligible minute in seven of eight later offline SEEG task tests sci:sha256:b75e8b3e5203a4d6767b14967d68fc454f0c807449693d88c83cfd3215bec19a

    This package supplied the eight continuous power records, original trial event arrays, contact selection files, source metadata, and pinned EDF range recipes. The new raw check adapts its EDF parsing and independent power calculation. All 72 source receipts matched the archive manifest.

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

    The analyzed source is OpenNeuro ds004473 version 1.0.2. Its version-pinned metadata records this DOI and a CC0 license.

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

Cited by

2 publications

Public catalog entries on Sciscape.