Rank bounds certified 79.7% of equally weighted state decisions in 28 SEEG patients against any finite change to one nonnegative shaft score with fixed calibration
With fixed calibration, exact rank bounds certified 79.7% of state decisions in the second selected SEEG record of 28 patients. Pre-onset and seizure windows received equal weight. The descriptive 95% patient interval was 76.0% to 83.5%. The bound permits any finite, nonnegative change to one shaft score. The test used public OpenNeuro ds004100 version 1.1.3 and 2,182 matched bipolar pairs outside onset shafts. A stable decision did not establish correct detection: 42.0% of seizure windows gave a certified no-alert decision. If only certified alerts were kept, seizure-window alerts fell from 47.1% to 35.2%. The package preserves the proof, reader code, all shaft vectors, source metadata and hashes, exact voltage audits, three figures, and repeat instructions. A fresh full EDF check reproduced all 19 retained pairs in one record. The study did not test stimulation or clinical control.
stereoelectroencephalography · brain-computer interface · median reader · rank bounds · decision stability · fault tolerance · seizure windows · calibration
References
2- usesLocal calibration lowered mean pre-onset window alerts from 25.6% to 8.8% for a median monitor in 28 SEEG record pairs after onset shafts were excluded sci:sha256:3f75c11b0a1b56a89e441827aad7b6b86812d0548cc966f8a1c9eea17cc7bd0e
We used its matched 28-patient SEEG record pairs, frozen powers and voltage audits, calibration rule, pair selection, source metadata and hashes, and ordinary median reader outputs. We added exact bounds for one changed shaft score and measured decision stability and retained alerts.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
Used the HUP iEEG Epilepsy Dataset, OpenNeuro ds004100 version 1.1.3, for the selected SEEG records and their source metadata. The recorded exact-version license is CC0.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesTargeted feature changes on one shaft caused wrong outputs from all 507 withheld windows in eight fixed SEEG output gates
Its median rank certificate provides context for shaft fault guarantees. It has a different decision rule and fixed calibration assumptions. Our result concerns the unanimity gate only.
- supportsMean AUC lower bounds exceeded 0.5 for site errors capped at 5% of training response SD in 20 iEEG patients and 20 fMRI adults
It separates exact rank decision stability from label accuracy in a different human SEEG task. We do not reuse its data or code.