SOZ label fraction showed no detected association with interictal HFA SOZ AUC across nine SEEG cohorts

We tested whether the fraction of implanted bipolar pairs labeled as seizure-onset zone (SOZ) changed how well interictal 80–150 Hz power (high-frequency activity, HFA) ranked SOZ pairs within patients. We reused an archived per-patient AUC table derived from four awake cohorts and five NREM/resected-zone cohorts; no new raw data were analyzed. The pre-specified statistic was the patient-weighted mean of within-cohort Spearman correlations between SOZ fraction and AUC, with within-cohort permutations, patient bootstrap confidence intervals, and Holm correction across four tests. For awake HFA (4 cohorts, 58 patients), weighted rho was −0.069 (95% CI −0.356 to +0.231; permutation p=0.62; Holm p=1.00). For HFA across all nine cohorts (121 patients), rho was −0.002 (95% CI −0.220 to +0.208; p=0.98; Holm p=1.00). Awake spike-rate and line-length tests also showed no detected association. Absolute SOZ-pair count and total pair count were not moderators (all p≥0.47). These results agree with a prior NREM-only analysis and do not support a SOZ-fraction adjustment at the reported precision. The all-cohort confidence interval excludes correlations beyond approximately ±0.22; the awake estimate is less precise. Limitations include small cohorts (6–30 patients), mostly low awake SOZ fractions, clinical rather than ground-truth labels, one clip per patient, differing source pipelines, and a between-patient design.

stereo-EEG · seizure-onset zone · high-frequency activity · AUC · epilepsy · interictal · cohort analysis