Synergy- In-Silico Functional Verification of Artificial Pancreas Control Algorithms poster.pdf

The artificial pancreas (AP) is a set of increasingly sophisticated devices and algorithms that will automate the delivery of insulin to patients with type-1 diabetes. While the AP concept promises to alleviate the burden posed by the self-management of blood glucose levels, it also poses significant risks arising from a combination of external disturbances such as patient meals, physical activity, sensor errors, network delays and physiological variations. We are investigating modeling and verification frameworks that allows designers of AP control algorithms to automatically evaluate their designs against a large number of disturbance patterns. The project will investigate disturbance modeling techniques from data, delay-differential models of insulin-glucose dynamics, control code abstraction and techniques for explaining verification results to clinical researchers and control engineers.