Raw delta–high-frequency coupling marked clinical SOZ pairs, but the effect was not significant after band-power control in ds003876

We tested competing findings about interictal phase-amplitude coupling (PAC) as a seizure-onset-zone (SOZ) marker in an independent four-centre SEEG cohort. From 30 interictal files in OpenNeuro ds003876 v1.0.2, we analyzed the first 60 s for each file. There were 2,668 bipolar pairs, including 218 SOZ pairs, from 29 patients with both SOZ and non-SOZ pairs. The plan was hashed before PAC values were computed. Raw Tort modulation index (3–4 Hz phase, 65–115 Hz amplitude) marked SOZ pairs (mean patient AUC 0.619, 95% CI 0.550–0.688; 21/29 patients above 0.5; Holm p=0.007). After control for 3–4 Hz and 65–115 Hz power, the effect was not significant (AUC 0.539, 95% CI 0.469–0.605; Holm p=0.58). After additional control for spike rate and contact index, AUC was 0.507 (95% CI 0.452–0.563; Holm p=0.83). These results replicate the raw PAC association but do not replicate the prior band-power-controlled claim; the fully controlled result agrees with a prior null finding. In a secondary analysis, 1–4 Hz mean vector length retained a small residual association after the full control (AUC 0.572, 95% CI 0.514–0.629; Holm p=0.040 in its test family), which needs confirmation. Band-power AUCs were 0.725 for 3–4 Hz and 0.681 for 65–115 Hz. Limits include one 60-s clip per patient, no sleep or artifact control, clinical SOZ labels, linear covariate control, contact number as a proxy for depth, 29 patients, and one UMF patient. Pair observations were not independent within patients; the patient was the test unit.

phase-amplitude coupling · intracranial EEG · seizure onset zone · epilepsy · replication · modulation index · mean vector length · OpenNeuro ds003876