The long term objective of diabetes treatment is to create an "artificial pancreas" to substitute for the role of the biological pancreas in patients. This artificial pancreas must constantly measure glucose and respond with an appropriate dose of insulin and/or glucagon to achieve an artificial analog to homeostasis. However, the artificial pancreas is subject to limitations not experienced in biological systems, and delivery of insulin by pumps has been known to be made inaccurate by factors such as the presence of magnetic fields, occlusions and disconnects in the lines, and even travel at high altitudes. As a result, patients can experience out-of-control glucose levels for hours before they or the artificial pancreas system becomes aware of a problem, and this results in degraded clinical outcomes. Insulin pumps are subject to such errors because of a simple but significant limitation - they cannot directly measure the rate of fluid flow. Pumps today estimate their rate of delivery by measuring processes internal to the pump, such as how many times a gear has turned, with the addition of separate sensors to deal with each of many possible faults. These indirect sensors cannot detect all possible variations in environment, however, and as a result closed loop control is temporarily lost. Fluonic has developed a simple, disposable flow sensor that can be integrated directly at the injection point at the patient, and which can record flow rates with 1% accuracy and provide immediate detection of infusion errors such as the growth of an occlusion. The long term goal of our study is to demonstrate the applicability of our flow measurement technology to ambulatory insulin delivery, so that it can be incorporated into artificial pancreas systems by pump manufacturers. In Phase I of this proposed work, we will demonstrate the applicability of our flow sensor for use in insulin delivery. We will use multiphysics modeling to optimize the design of our devices for the unique flow characteristics of ambulatory insulin pumps, so that a single chip can accurately measure both low basal flow rates and the higher flow rates seen during bolus administration. We will further optimize the design to minimize power consumption, in order to demonstrate compatibility of the sensor with the portable battery requirements of an ambulatory artificial pancreas. Finally, we will demonstrate the successful design phase by measuring flow created by an off-the-shelf ambulatory pump in both basal in bolus mode in the laboratory over a three day period, and demonstrating rapid response to the presence of an artificially-produced occlusion. The Phase I program will culminate in a design for a portable sensor unit, which we will build and test in Phase II.
The development of flow monitoring for ambulatory insulin pumps will enable instant detection of flow error such as infusion line faults, and overall improved operation of AP algorithms for medical infusion. This will reduce the time a diabetes patient spends with out of control glucose levels, resulting in improved clinical outcomes.