In 20 SEEG patients, a causal 250–500 Hz veto removed 37% of fast-ripple-marked triggers and 2.7% of pure-ripple triggers

We assessed whether a causal 80–250 Hz ripple trigger also fires on author-marked fast-ripple (FR; 250–500 Hz) events, and whether a causal FR-band envelope at the trigger sample can veto such triggers. Using 120 s from run-01 during interictal NREM sleep for each of 20 patients with epilepsy in OpenNeuro ds003498, we evaluated 481 bipolar pairs and 66,619 held-out triggers. A median 8.8% [5.3, 15.7] of marked triggers per patient were FR-type; 9/20 patients had at least 10%. In 18 patients, the causal FR-band feature separated FR-type from pure-ripple triggers with median AUC 0.78 [0.67, 0.85] (18/18 above 0.5; p = 7.6e-6). A veto at robust z > 3 removed a median 37% [17, 58] of FR-type triggers and 2.7% [1.7, 4.8] of pure-ripple triggers, with no added delay. The reference FR marks were produced by an automatic detector that also uses FR-band energy; these results measure agreement with those marks and do not establish that vetoed events are pathological. The analysis used one 120 s run per patient, assumed marks were complete on marked pairs, and did not check overlap of `frandr` and `ripple` marks.

fast ripples · hippocampal ripples · closed-loop stimulation · intracranial EEG · causal detection · veto · NREM sleep · epilepsy