Hypoglycaemia-Related EEG Changes Assessed by Approximate Entropy

Several studies performed in human beings demonstrated that glucose concentration in blood can affect EEG rhythms, typically evaluated by standard spectral analysis techniques. In the present work, we investigate if EEG complexity assessed by a nonlinear algorithm, Approximate Entropy (ApEn), reflects changes of glucose concentration levels during an induced hypoglycaemia experiment. In particular, in 10 type-1 diabetic volunteers, ApEn was computed from the P3-C3 EEG channel at different temporal scales and then correlated to the three classes of glycaemic states, i.e. hyper/eu/hypo-glycaemia. Results show that, for all considered temporal scales, EEG complexity in hypoglycaemia is lower, with statistical significance, than in eu- and in hyperglycaemia. No statistically significant difference can be evidenced between ApEn values in hyper- and in eu-glycaemic states. In conclusion, in addition to power indexes in the four traditional EEG bands, other indicators, and ApEn in particular, can be used to quantitatively investigate glucose-related EEG changes.