On-line glucose prediction from continuous monitoring subcutaneous sensor

Abstract: A clinically important task in diabetes management is the prevention of hypo-(and hyper-) glycemic events. In this work, we assess the feasibility of approaching the problem by exploiting recently developed continuous glucose monitoring (CGM) devices. In particular, we study the possibility to predict ahead in time glucose levels by exploiting their recent history monitored every 3 min by a minimally invasive CGM system, the Glucoday, in 13 diabetic volunteers for 48 h. A simple prediction strategy, potentially usable on-line, is assessed by considering two different prediction horizons, i.e. 30 and 45 min, which can have practical relevance. The prediction performance is evaluated by assessing the delays with which significant points of the original glucose time-series (major peaks and nadirs and critical threshold crossings) can be detected from the predicted time-series. Our results show that these delays are always, on average, significantly lower than the prediction horizon, suggesting that hypoglycemic/hyperglycemic states can be forecasted in advance on the basis of the data provided by GCM systems.