Pseudo real-time Diabetes Detection & Analysis using EEG signals and implementation on Raspberry Pi

Abstract

Diabetes is a worldwide cause of major diseases, related to gastrointestinal tracts and high sugar levels. Diabetes causes the body to produce insufficient amounts of insulin (responsible for keeping the blood sugar levels in body in check) and therefore, resulting in uncontrollable sugar levels, that lead to many complications in the human body. The aim of this project was to perform data analysis of the Eps (Evoked Potentials) to determine if there are differences in the way blood glucose levels affect the EEG in diabetic patients and healthy control subjects, and how. This in turn has an adverse effect on parts of the human brain, such as the cerebellum, hypothalamus etc. This also effects the performance of the brain, as it’s not able to perform its normal functions, the EEG’s clearly show us the brain scans of a normal person and the ones of a diabetic patient. Prolonged effect of diabetes may cause heavy brain damage and in severe cases brain death. Therefore the main working of our project is to detect and then analyze the EEG signals of normal (control) patients and those of diabetic patients using features that give us the best accuracies and can accurately tell which EEG signal belongs to which subject (health or patient).. The resulting signals processed by the raspberry pi will give us information about the signals, their features such as Skewness, mean, range, moment etc., and then using these features the Pi will be to accurately tell which EEG belongs to the patient or the healthy subject. The Pi acts as the server side and the PC acts as the client side, sending EEG signals to the Pi.