Automated glucose control in the ICU: effect of nutritional protocol and measurement error

Tight glycaemic control has been shown to reduce mortality and morbidity in critically ill subjects. Using in silico computational approach, the objective of this study was to evaluate the effect of nutrition and the measurement error on glucose control. In silico simulation environment describing 21 synthetic subjects was used to simulate a 48 h clinical trial with an adaptive model predictive controller in the intensive care unit. Two types of nutritional protocols, simple and complex, and various levels of the measurement error (ME) were evaluated. The simple nutritional protocol resulted in more efficacious glucose control compared to that obtained with the complex nutritional protocol. A considerable deterioration was noted with the increasing level of the ME. Severe hypoglycaemia episodes (<2.8 mM) were observed with the ME>10%. We conclude that nutritional protocol should be kept simple to facilitate efficacious glucose control with an adaptive model predictive controller. The measurement error of the glucose measuring device should be less or equal to 10%