Bio-inspired systems for treatment of diabetes

Abstract

There have been several technological advancements aimed at improving glycaemic control in people with Type 1 diabetes such as continuous glucose monitors (CGM) and continuous subcutaneous insulin infusion pumps (CSII). However, even with these advances, the millions of people living with diabetes struggle to keep their glucose levels within target range (4-8 mM). There are various reasons for this sub-optimal control of glucose levels which include lack of patient compliance, and most importantly, determining the amount of insulin needed to maintain glycaemic control is not a trivial task since glucose levels are affected by many different parameters. Since a healthy pancreas maintains optimal glycaemic control, a bio-inspired system that replicates the natural mechanisms of glucose sensing and physiological insulin secretion would be ideal for maintaining glycaemic control in people with Type 1 diabetes. Towards this aim, in this thesis a Bio-inspired Artificial Pancreas (BiAP) has been designed and implemented. It runs on dedicated hardware, and incorporates a bio-inspired controller that replicates the insulin-secreting physiology of the pancreatic β-cells. The controller runs on an embedded microchip and the system has been optimized for low power operation. In addition a glucagon controller is used with the aim of reducing hypoglycaemic episodes. The low-power, hand-held unit (BiAP) interfaces to a continuous glucose sensor (Medtronic Enlite or Dexcom G4) and sends the required insulin and glucagon dose information to two subcutaneous pumps (Roche Accu-Chek Combo). The BiAP has undergone over 250 hours of closed loop studies at the NIHR/Wellcome Trust clinical research facility at Hammersmith hospital. The results show that the BiAP is capable of achieving good control with both the single and dual hormone systems. Additionally, the hand-held unit achieves power efficient operation, lasting for 54 hours during full control. Furthermore, it was identified that one of the challenges to the successful realisation of an artificial pancreas system is sensor accuracy. Therefore, a bio-inspired methodology for improving glucose sensor performance is presented. Specifically, the noise shaping capability of gap junction coupling is utilised to synchronise β-cell behaviour to a varying glucose sensor input and reduce random noise. In addition, a CMOS chip was fabricated containing 4 β-cell circuits, which were then coupled off chip. These improvements will allow more accurate glucose sensing for use in an artificial pancreas.