In 28 HUP SEEG patients, leading spike events marked SOZ pairs less well than spike rate (AUC 0.68 vs 0.76; Holm p < 0.001), while event order did not (AUC 0.52)

In short interictal SEEG clips from 28 patients in OpenNeuro ds004100, we tested whether bipolar pairs that led multi-channel spike events identified clinical seizure-onset-zone (SOZ) pairs better than spike rate. Spike detections used a 10–60 Hz amplitude threshold of 6 robust sigma; event leadership was based on the peak time of each pair’s filtered signal. Mean SOZ classification AUC was 0.757 for spike rate and 0.681 for leader rate. The paired leader-minus-spike AUC difference was −0.076 (95% CI −0.106 to −0.046; worse in 20/28 patients; Holm-adjusted sign-flip p = 9 × 10⁻⁵). Earliness, the mean rank of a pair within events and independent of event participation count, had mean AUC 0.525 (95% CI 0.436 to 0.611; n = 18; p = 0.60). Participation rate had mean AUC 0.761. The same direction for the primary comparison appeared across four event-definition sensitivity settings. These results indicate that, for this detector and these short clips, SOZ information in leader rate was associated with how often a pair spiked, while within-event order did not provide a useful SOZ marker. The result is limited by one short interictal run per patient, an amplitude detector without expert review, peak timing in bipolar signals, and clinical SOZ labels as the reference.

stereo-EEG · interictal spikes · seizure-onset zone · epilepsy · event timing · OpenNeuro · AUC