The Artificial Pancreas

Abstract

An artificial pancreas is an engineered device for treating diabetes, which is one of the greatest epidemic diseases in the US and worldwide. In the US 105 million people have diabetes or prediabetes. An artificial pancreas is a device containing only synthetic materials which substitutes for an endocrine pancreas by sensing the BG level, determining the amount of insulin needed, and then delivering an appropriate amount of insulin. An artificial pancreas uses closed loop control by measuring glucose continuously and then continuously adjusting an infused dose of insulin. The three main components of an artificial pancreas are: 1) a continuous glucose sensor; 2) an insulin infusion pump; and 3) a controller using software to determine the insulin dose. Wireless radios link the components. Better sensors, better insulin delivery, and better control algorithms are all needed to create a viable system. Improvements in all three components are currently being developed. Control algorithms typically use either a Proportionate/Integral/Derivative method or a Model Predictive Control method. Each type of control formula has advantages and disadvantages. A controversy in control of a closed loop system involve how sensitively to program the controller to deliver a meal bolus of insulin in case of a rise in blood glucose, which could represent either a meal requiring insulin right away or a random fluctuation requiring no additional insulin. A second controversy involves how to protect a patient from hypoglycemia, because insulin lowers elevated glucose levels, but no part of a classical artificial pancreas system can raise depressed glucose levels. Glucagon is being tested as a rescue treatment for hypoglycemia in closed loop control systems. Current closed loop systems have already been demonstrated to be effective in specified closely monitored inpatient environments. Emerging and future closed loop glycemia maintenance systems will incorporate many inputs besides glucose levels. These inputs will be necessary to use much more complex models of glycemia which can be nuanced by such factors as day-to-day insulin sensitivity, meals, exercise, stress, and individual differences in the time course of insulin action. In the future, an artificial pancreas system will provide telemedicine care in case of hypoglycemic emergencies and will likely offer an option for a remote healthcare provider to assume control of the insulin programming to override sensor-determined control. Recent reports and meetings by FDA, Diabetes Technology Society, NIH, and Juvenile Diabetes Research Foundation are intended to overcome barriers to designing robust clinical trials of artificial pancreas systems. The first artificial pancreas product is approved in Europe and is at an early stage of undergoing testing in the United States. This product contains a low glucose suspend feature in the event of a hypoglycemic episode, which fails to result in acknowledgement of the glucose level or adjustment of the insulin dose. At this point in order to see a commercial closed loop product be approved for marketing and reimbursement, we will need to collect additional data about the safety, efficacy, and economic impact of these systems. A well designed artificial pancreas system will revolutionize diabetes care.