A fixed pulse limit increased detection of imposed SEEG gain trends in eight recordings under an exchangeable pulse model

We tested a sequential rank alarm on public SEEG responses from eight people with epilepsy. The data contained 85 sites, 879 bipolar readouts, and 987 pulses. The primary alarm used two pilot pulses, stake 0.5, and threshold 20. We proved that it needs at least 13 total pulses. It could not fire at 69 sites within the recorded count. In paired software tests, it detected a gradual 50% recording gain decrease in 0.65% of trials. A secondary rule set its threshold for a fixed pulse limit with 65,535 null rank paths. It detected the same decrease in 41.08% of trials, with a descriptive person bootstrap interval from 30.81% to 51.24%. Its unchanged-gain reference alarm rate was 5.06%. The error proof requires pulse exchangeability conditional on the pilot and fixed readouts. The fixed limit must be declared before monitoring. Recorded changes had no validated labels. Constant positive gain left the ranks unchanged. The local historical archive preserves code; this SCI package preserves compact data, five figures, source hashes, all results, and retrieval instructions. A fresh full-source check for one person reproduced every matrix and table exactly.

sequential rank test · exchangeability · change detection · stereoelectroencephalography · pulse limit · signal monitoring