Hyperglycaemia is commonly found in critically ill patients. Clinical studies demonstrated that tight blood glucose control in medical and surgical ICU patients results in a significant better outcome for the patients. Based on this emerging clinical evidence, there are increasing efforts worldwide to maintain strict glycaemic control in critically ill patients. However, achieving this goal requires extensive nursing efforts, including frequent bedside glucose monitoring and the implementation of complex intensive insulin infusion protocols. A fully automated algorithm may help to overcome some of these limitations by excluding intuitive interventions and integrating relevant clinical data in the decision-making process. This study will investigate the performance of an eMPC algorithm adjusted to target the range 4.4 - 8.3 mmol/L in line with the Surviving Sepsis guidelines.

Inclusion Criteria:

• Age: > 18 years of age
• Admitted following cardiac surgery
• Stay in the ICU expected to be > 20h
• Blood glucose > 6.7 mmol/l within 4 hours of admission to intensive care or patient already receiving insulin treatment

Exclusion Criteria:

• Patients with hyperglycaemic crisis/ketoacidosis due to insulin deficiency.
• Known or suspected allergy to insulin
• Any disease or condition which the investigator or treating physician feels would interfere with the trial or the safety of the patient (i.e. liver failure, other fatal organ failures)
• Patients participating in another study
• Moribund patients likely to die within 24 hours
• Patients after organ transplantation within the last three months
• Patients under high dose cortisol treatment (cortisol > 1000 mg/day or equivalent doses of hydrocortisol)