Abstract

A clinically important task in diabetes management is the prevention of hypo/hyperglycemic events. The availability of continuous glucose monitoring (CGM) devices allow to develop new strategies, but new problems have also emerged. In this contribution, we discuss three major challenges which, in practical real time CGM applications, should be dealt with: filtering to enhance the signal-to-noise ratio, ahead-of-time prediction of glucose concentration, and generation of hypo/hyper-alerts. For all these challenges, some techniques, with a different degree of sophistication, have been proposed recently in the literature, but several issues remain open.

Keywords: Diabetes, Prediction, Filtering, Polynomial model, Auto-regressive model, Alert