In 18 HUP SEEG patients, avalanche exponents fell with bin width and remained near 1.5 after pair timing was removed

We analyzed 120-s interictal run-01 clips from the first 18 of 38 HUP SEEG patients whose data downloaded completely. For bipolar 1–45 Hz local field potential events thresholded at |z| > 3, the avalanche size exponent at the mean pooled inter-event interval (median 21 ms) was 1.507 [1.469, 1.526], not different from 1.5 (p = 0.47). Across 2, 4, 8, and 16 ms bins, alpha was 2.86, 2.37, 2.00, and 1.69, decreasing by 0.39 [-0.43, -0.37] per bin doubling (p = 5e-5). Independent circular shifts that preserved each pair's event train but removed timing between pairs gave surrogate alpha 1.436; real alpha exceeded surrogate alpha by 0.062 [0.049, 0.089] (p = 5e-5). Real avalanches were larger than surrogate avalanches (18/18 patients, Holm p = 1.5e-4). The multistep-regression estimate was m = 0.986 and tau = 274 ms, while the fit grid guaranteed m < 1; real m was lower than surrogate m. SOZ participation in large avalanches was not higher (n = 17, p = 0.11). These results show that, in these clips, alpha near 1.5 was not a specific sign of criticality and depended on bin width and event rates. The sample was nonrandom and limited to one short interictal clip per patient; events were LFP threshold crossings, not neuronal spikes.

neuronal avalanches · intracranial EEG · SEEG · bin width · criticality · surrogate analysis · epilepsy · multistep regression