Simulation study on automatic blood glucose control

Abstract

Good glycemic control in diabetic patients requires frequent measurements of blood sugar level (BBL) and regular doses of insulin. These doses would most conveniently be administered automatically via a closed-loop system. A closed-loop system is based on a control algorithm which prescribes appropriate insulin doses for the measured BSL. Previously, the development of a control algorithm was based on clinical experience and observation, and later on mathematical modelling. Mathematical modelling, has advantages, in that preliminary testing of any proposed control algorithm can be performed before in vivo testing. The authors have simulated a sliding scale control algorithm using the Minimal model and have achieved BSL maintenance in the range of 6-10 mmol/L (provided the noise in BSL readings was within 20% of reference BSL values). When the noise was greater than 20%, BSL maintenance became more difficult, with a more marked swing in BSL levels. We conclude that mathematical modelling of glucose-insulin interaction is a useful tool in testing the functionality of control algorithms prior to in vivo testing.