In HUP SEEG, seizure-onset map drift was not significant, but recent seizures predicted later onset pairs better than the first

In 17 patients with 71 seizures from one center, similarity of bipolar seizure-onset maps tended to fall with run-order lag, but the pre-specified permutation test was not significant for 20–100 Hz power (mean patient r = -0.23, p = 0.072) or line length (r = -0.14, p = 0.28). A pre-onset background map did drift (r = -0.30, p = 0.015); after regressing out background similarity, onset drift was smaller and not significant (r = -0.15, p = 0.24). For 37 target seizures from patients with at least three seizures, the most recent seizure predicted the next seizure’s top 10% high-frequency onset pairs better than the first seizure (median patient AUC 0.85 vs 0.69; 11/17 patients higher; exact Wilcoxon p = 0.023). The mean of all prior seizures performed similarly to the most recent seizure (0.87 vs 0.85, p = 0.75) and better than the first (14/17, p = 0.003). The first-seizure deficit was concentrated in about four patients with atypical first maps, rather than showing slow drift in every patient. The results support updating onset targets with each new seizure, while the primary drift result remains inconclusive. This single-center study assumes run order is seizure order; actual intervals are unknown, and analyzed lags were 1–4.

seizure onset · stereo-EEG · seizure mapping · high-frequency power · line length · closed-loop stimulation