A count gate limited output to six commands per rolling minute and kept 19 hits in eight offline SEEG tests

We tested whether an early score threshold kept a later rolling command limit. We used eight public SEEG streams from OpenNeuro ds004473 version 1.0.2. The early threshold met a limit of six commands per rolling minute in calibration. It exceeded that limit in one later test. A causal count gate met the limit in all eight tests. It reduced output from 85 to 83 commands and kept the same 19 hits among 340 keypresses. At a fixed limit of three commands, it reduced hits from 15 to 13. The package gives a proof, saved-state gate code, all command times, three figures, and checks of all continuous features from raw data. The count proof needs a complete output history and a monotone clock. Detection was low. These offline results do not establish a safe stimulation dose or live BCI benefit.

stereoelectroencephalography · brain-computer interface · rolling command limit · count gate · causal control · offline benchmark