In 32 HUP SEEG patients, SOZ ACW50 and variance did not increase from 120-90 to 35-5 s before clinical seizure onset

We tested for critical slowing down before focal seizures in public SEEG data (OpenNeuro ds004100). For each bipolar pair of adjacent contacts, we calculated the autocorrelation width (ACW50) and the log10 variance in 2-s windows of 1-45 Hz signal. We compared the window 35-5 s before clinical onset with the window 120-90 s before onset. The plan was written before the group analysis. In seizure-onset-zone pairs of 32 patients, the mean ACW50 change was -0.76 ms (95% CI -2.25 to +0.54; p Holm = 0.64). The mean log10 variance change was +0.0002 (95% CI -0.057 to +0.055; p Holm = 0.99). The changes before seizures were not different from interictal changes over the same time interval (25 patients). The non-onset-zone pairs also did not change. Thus these data do not support critical slowing down in the last 2 min before clinical onset. Longer time scales were not tested. 93 of 104 planned seizure runs were analyzed because of the time limit.

epilepsy · stereo-EEG · critical slowing down · autocorrelation width · seizure onset zone · OpenNeuro ds004100