Detection of hypoglycemia using measures of EEG complexity in type 1 diabetes patients

Abstract: Previous literature demonstrated that hypoglycemic events in patients with type 1 diabetes (T1D) are associated with measurable electroencephalography (EEG) changes in power spectral density. By using a dataset of 19-channel EEG recordings in 34 hospitalized patients with T1D who underwent a hyperinsulinemic–hypoglycemic clamp study, in the present work we show that hypoglycemic events are also characterized by EEG complexity changes quantifiable at single-channel level by empirical conditional and permutation entropy, and fractal dimension indices, i.e., Higuchi index, residuals and tortuosity. Moreover, we demonstrate that the EEG complexity indices computed in parallel in more than one channel can be used as input of a Neural Network aimed at identifying hypoglycemia and euglycemia. The achieved accuracy of about 90% suggests that nonlinear indices applied to EEG signals might be useful in revealing hypoglycemia events from EEG recordings in T1D.