Résumé
We investigated the feasibility of closed loop insulin delivery by using a model based upon the coupling of an implanted insulin pump that infuses insulin by intra-peritoneal route and an intravenous enzymatic glucose sensor. Closed Loop Control (CLC) by a proportional, integral and derivative algorithm was initiated in 4 type 1 diabetic subjects, already implanted with both devices (MiniMed-Medtronic, Northridge, CA, USA), and continued for 48 hours. Three meals of 40 g and twice 80 g carbohydrates were served daily. Blood glucose was assessed every 10 minutes for the first two hours of meals and every 30 minutes otherwise. Algorithm parameters were empirically adjusted during the experiment based on antecedent blood glucose levels. CLC in one subject was stopped after 24 hours due to inadequate sensor performance. During CLC, the algorithm kept glucose within 80–240 mg/dl for 84.1% of the time. Algorithm retuning did not change the percentage of glucose >240 mg/dl but increased the percentage within the 80–120 mg/dl range during the final 24 hours. Excluding meals, glucose was <240 mg/dl for 98% of the time. We conclude that once tuning optimization is achieved, automated control of blood glucose is feasible using this model. Optimal closed loop insulin delivery compensations remain to be defined to allow permanent restoration of normal glucose control.