Abstract

The main risk for patients with type 1 diabetes (T1D) is to fall into hypoglycemia. We have extended the quantitative detection of hypoglycemia from the altered EEG signal in T1D patients by analyzing all EEG channel data through different measures of signal complexity such as the fractal domain and entropy indices. Finally, they were classified through a neural network in order to detect hypoglycemia with a high percentage of precision using the results obtained from the complexity analysis.