Hypoglycemia-Induced EEG Changes in Type 1 Diabetic Subjects

Background: Hypoglycemic events have been proven to be associated with EEG changes, especially in the low frequency bands, suggesting the possible role of the brain as a biosensor to detect hypoglycemia in real-time. Many indices can be extracted from the EEG, in particular in time, frequency, and time-frequency domains, but the set of the most sensitive to hypoglycemia is not completely established. Methods: EEG recordings and sparse blood glucose (BG) concentrations were collected in parallel in 18 T1D subjects during an insulin-induced hypoglycemia experiment. P3-C3 (P4-C4) and P3-T3 (P4-T4) EEG recordings were assessed by linear spectral analysis (in canonical as well as in individualized bands), variability of EEG power modulation, and nonlinear complexity indices. Statistical significance of the changes of EEG indices during the transition from eu- to hypo-glycemic conditions has been evaluated. Results: In all the domains of analysis, statically significant differences in the EEG signal by passing from eu- to hypo-glycemia have been observed. For instance, an increase of the power spectral density in both theta and alpha bands (in particular in the left brain channels), a significant decrease in EEG complexity measured by Approximate Entropy, and an increase of the variability of the reactivity index in theta band were noted. Conclusion: Remarkable changes of some EEG indicators measurable in real-time in time, frequency, and time-frequency domains have been shown to occur during insulin-induced hypoglycemia. Possible use of these indicators in the real-time detection of hypo-events will be a matter of future investigations.