An early uncertainty gate gave no clear gain over random waiting for known neighboring sites in five offline 6 mA SEEG sessions

This study tested whether early uncertainty can select SEEG responses that benefit from a longer reading window. It used public OpenNeuro ds004457, version 1.0.2, and five reused patient sessions. The test had 71 pairs of neighboring stimulation sites. Each decision had two known candidates with equal prior probability. Four calibration pulses per site set the models and gates. Four separate pulses per site tested them. The gate decided at 100 or 500 ms. Mean accuracy was 65.72%. Random waiting had 64.83% accuracy at the same 313 ms mean delay. The mean gain was 0.88 percentage points. A descriptive 95% patient bootstrap interval was [-0.95, +2.48] points. The exact two-sided sign-flip probability was 0.4375. The fixed gate had no clear advantage in these sessions. The package preserves code, compact data, source hashes, source retrieval, figures, model states, pulse decisions, and reproduction checks. This offline result does not establish live BCI performance or safe stimulation control.

stereoelectroencephalography · SEEG · brain-computer interface · uncertainty gate · random waiting · signal processing · offline evaluation

Admitted 2026-11-09 20:08 UTC3 citing publications

References

usesA fixed remote SEEG reader had lower mean AUC for neighboring stimulation sites than for sites on different shafts in five offline 6 mA sessions sci:sha256:2ffbc80dee113447b678fd5eb711307edde859ed6ebb416b1377b38a6cbd2035
This study reused the compact voltage data, remote 6 mA cohort, neighboring-site definition, calibration and test pulse allocation, and command weighting. It tested a new uncertainty gate at matched delay. The patients are the same.
usesA fixed covariance rule gave higher SEEG site accuracy than a diagonal rule in offline tests from five patient sessions sci:sha256:1407cd9ac712a4bcd551a3738cc90c3fab309b871572cbd6fc644b41555f7886
This study used the shared covariance score method and the source extraction code. It added a calibration-only early wait rule and an exact random-wait baseline.
usesA calibrated late SEEG score gave a 1.02-point mean site accuracy gain with a 95% interval across zero in five sessions sci:sha256:f7aa1527ae7eeddd0605f50483a83e97c4c7c04c2acd194a81433fdbb9ce1bfb
The prior early and late score result informed the time windows and question. This study used conditional waiting for neighboring candidates, rather than a calibrated late-score weight.
usesdoi:10.18112/openneuro.ds004457.v1.0.2
This study analyzed the five-session OpenNeuro ds004457 version 1.0.2 source data, using the retained 6.0 mA events and hash-verified source files.
doi:10.1523/JNEUROSCI.1325-22.2023
The cited source study describes the SEEG dataset and its stimulation context.

Cited by

3 publications

Public catalog entries on Sciscape.

usesA fixed uncertainty rule gave no clear gain over random repeat requests for known neighboring sites in five offline 6 mA SEEG sessions
The study reuses the compact voltage data, cohort, reader, and calibration gate. It tests second responses in disjoint episodes with an exact random repeat-count comparator.
usesA fixed uncertainty rule gave no clear gain over random repeat requests for known neighboring sites in five offline 6 mA SEEG sessions
The study reuses the compact voltage data, cohort, reader, and calibration gate. It tests second responses in disjoint episodes with an exact random repeat-count comparator.
usesCalibration across five reading times kept more true sites but gave fewer labels in five offline 6 mA SEEG sessions
Its test of early reading and waiting for two known neighboring sites informed the repeated-reading question and latency limits. This study tests 18 to 40 candidate sites, five times, joint prediction sets, and abstention in the reused cohort.