Spike rate and high-frequency activity did not show patient-level complementarity in 39 NREM SEEG patients

A pre-registered replication tested whether interictal spike rate and high-frequency activity (HFA) fail to identify the clinical seizure-onset zone (SOZ) in different patients. We analyzed archived per-pair features from 90-s NREM sleep clips for 39 patients with at least two SOZ and two non-SOZ pairs (699 pairs, 156 SOZ), using OpenNeuro ds007118, ds007119, and ds007120 (version 1.0.0). Across patients, the Spearman correlation between spike-rate and 80–300 Hz log-power SOZ AUCs was 0.33; the shared-label permutation null mean was 0.31 (95% interval 0.01–0.58; lower-tail p=0.54; Holm p=0.85). The pre-registered complementarity hint from an earlier awake cohort therefore did not replicate at this sample size. Spike rate versus slow-wave-HFA modulation index had rho=0.42 (null mean 0.43; Holm p=0.85). Combining spike rate and HFA by within-patient rank mean changed mean AUC by +0.031 relative to spike rate alone (95% CI −0.022 to +0.084; Holm p=0.80). All four feature-pair correlations were within 0.04 of their permutation-null means. The study has limited power to detect the earlier effect, uses clinical SOZ labels and short NREM clips, and relies on an automatic spike detector. The null assumes label-sampling covariance is the same with and without true signal; no shaft-block sham check was performed. Inputs, code, pre-registration, results, and reproduction steps are included.

stereo-EEG · seizure-onset zone · interictal spikes · high-frequency activity · NREM sleep · replication · permutation test · AUC