SEEG recruitment order repeats across seizures in 16 patients, but excess similarity is not significant among jointly recruited pairs

We compared seizure-to-seizure recruitment latencies in 16 patients from OpenNeuro ds007445 v1.0.4, using the first three seizure sessions and 1-s log line-length latencies. Across 38 seizure pairs, mean Spearman similarity was 0.46 (95% CI 0.28 to 0.61), and similarity beyond a within-region shuffle null was 0.30 (95% CI 0.17 to 0.42; Holm-adjusted p = 0.001). The result also held outside the seizure onset zone. Thalamic recruitment rank was patient-specific (ICC(1) = 0.65, permutation p = 0.0002). However, when restricted to pairs recruited in both seizures, overall similarity was 0.30 (p = 0.018), while excess similarity beyond region classes was not significant (0.14, 95% CI -0.03 to 0.30; p = 0.12). Much of the primary effect reflects which pairs were recruited within 60 s; about half of values were censored ties. These results support repeated spatial extent of spread, but only weakly support repeated fine timing order among recruited pairs at 1-s resolution. Seizures were not matched by type, spatial proximity was not controlled, and results come from one dataset.

SEEG · seizure · recruitment latency · stereotypy · thalamus · intracranial EEG · Spearman correlation