Type 1 diabetes mellitus is a disease characterized by complete pancreatic Beta-cell insufficiency. The only treatment is with subcutaneous or intravenous insulin injections, traditionally administered in an open-loop manner. Patients attempt to mimic normal physiology in order to prevent the complications of hyper- and hypoglycemia (elevated glucose levels, and low glucose levels, respectively). Even minor glucose elevations increase the risk of complications (retinopathy, nephropathy, and peripheral vascular disease).

In recent years, sensors and pumps have become available that show great promise for a closed-loop artificial pnacreas -- however the crucial missing component is the algorithm to connect the devices. In order to normalize the glucose levels of insulin dependent, type 1 diabetic patients, the algorithms for the development of an artificial pancreatic islet need to exploit all the measured variables that the normal islet insulin secretion utilizes and quickly increase or decrease the insulin secretory.

Our group has been working on model-based control algorithms for pump control over the last 15 years; with clinical evaluations over the last 6 years in collaboration with the Sansum Diabetes Research Institute. Our investigations have addressed the critical algorithmic elements of: model identification, disturbance estimation, model predictive controller design, event detection, monitoring & alarming, and optimization solution. In this talk, we present our most recent computational and clinical results in pursuit of the artificial beta cell. Our novel contributions include the model formulation, meal detection & estimation schemes, efficient programming formulation, and the use of insulin-on-board constraints to ensure safety.