Six checks gave no clear AUC gain over one 100 ms check in eight SEEG sessions after eight calibration pulses
We tested six response checks against one check at 100 ms in public SEEG data from eight epilepsy patients. SEEG means stereoelectroencephalography. We kept 86 readouts fixed. At each site, five pulses fit the rule. Three later before-pulse windows set a separate threshold. The remaining 354 pulses tested both rules. The endpoints were 30, 40, 50, 60, 75, and 100 ms. Each new score used only samples before its endpoint. Mean AUC was 0.936 for six checks and 0.955 for one check. AUC measures the ordering of after-pulse and before-pulse scores. The paired difference was -1.87 percentage points. The descriptive 95 percent person bootstrap interval was -4.23 to 0.79 points. The exact person sign-flip test gave p = 0.21875. Six checks gave no clear AUC gain. Their mean capped wait was 41.1 ms, but they called 29.8 percent of before-pulse windows. Reusing the single-check threshold raised that rate to 54.8 percent. With three separate controls, the maximum-control rule gives a 25 percent marginal false-call bound only under exchangeability. This offline test does not establish an online neural response detector. The package preserves data, code, source hashes, fits, scores, figures, and instructions.
stereoelectroencephalography · electrical stimulation · prefix scoring · AUC · calibration · brain-computer interface
References
5- usesAn AR(16) model with a 1 percent floor gave no clear 20–500 ms SEEG detection gain after five calibration pulses in eight epilepsy sessions sci:sha256:f498186f5bf9aa390e84158392afda66f1813af870a8cf045712bf3db794d97b
This publication supplied all 1042 good-pulse voltage windows at 86 fixed readouts. It also supplied metadata, hashes, retrieval code, and inherited audits. We used these inputs to test six prefix checks with separate control thresholds.
- usesAfter five calibration pulses, a 100 ms endpoint gave mean later AUC of 0.951 for fixed SEEG pairs in eight sessions sci:sha256:bfd74ee6170a5883952f42e0898b29d958be62240dad7866aafd0c75d34aa016
This publication tested fixed observation endpoints and motivated the 100 ms comparator. Our study tests a running maximum score and uses a different split with three separate threshold controls. We do not claim to reproduce its numerical result.
- usesFive calibration pulses gave mean later signed template AUC of 0.939 compared with 0.859 for one pulse in eight SEEG sessions sci:sha256:1d2aa103275ad7400befec4c899740ede8f2eb59a09897bdc3f4d4e190e54418
This publication defined the first-five energy selection rule and the fixed readout choices inherited through the direct input. We kept those choices and checked the stored budget-choice union. We did not repeat its full candidate search.
- usesdoi:10.18112/openneuro.ds004696.v1.0.1
The analysis uses the version 1.0.1 source windows and metadata from OpenNeuro ds004696.
- doi:10.1523/JNEUROSCI.2201-22.2023
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesThresholds from other patients gave a 10.2% mean background call rate at a 5% target after five fit pulses at 86 selected SEEG sites in eight patients
Used the five-fit, three-background-calibration, later-test split and the local background score standardization as the basis for a different test of threshold transfer between patients. Its repeated-check results informed the choice of question.