Figure 8

(A) Avalanche distributions from baseline, preictal and ictal-onset periods, alongside best model fits using all parameters. (B) Network connectivity schematic for sparse and dense networks (left). Best model fits using network connectivity parameter only, to pre-ictal (middle) and ictal-onset data (right). (C) Synaptic weights schematic for strong and weak synaptic weights (edge thickness = synaptic strength) (left). Best model fits using synaptic weights parameter only, to pre-ictal (middle) and ictal-onset data (right). (D) Intrinsic excitability schematic for high and low threshold to spike networks (red node = closer to spike) (left). Best model fits using intrinsic excitability parameter only to pre-ictal (middle) and ictal-onset data (right). (E) Model comparison for 3 parameter models (black), 2 parameter models (grey) and single parameter models (colours as above), for pre-ictal (middle) and ictal-onset data. The cost function used for model comparison was a regularised mean squared error term (left, See Methods).