Predicting blood glucose levels in diabetics using feature extraction and Artificial Neural Networks

Abstract

Diabetes mellitus is one of the most common chronic diseases. The number of cases of diabetes in the world is likely to increase more than two fold in the next 30 years; from 115 million in 2000 to 284 million in 2030. In type I diabetes, the disease is caused by the failure of the pancreas to produce a sufficient amount of insulin which leads to an uncontrolled increase in blood glucose unless the patient administers insulin, typically by subcutaneous injection. This work is concerned with managing diabetic patients by trying to predict their glucose levels in the near future (30 minutes) on the basis of the current levels. The goal of this paper is to determine the future blood glucose value of a diabetic patient using an Artificial Neural Network. Unlike other approaches, which involve questioning the patients, feature extraction procedures were implemented (Diabetic Dynamic Model) on diabetic blood glucose time series, in order to extract a knowledge (how patient blood glucose level will change a according to the external food intake and any human day activities). An Artificial Neural Network was then trained using these features in order to predict the future value of blood glucose level with an accepted accuracy.