The Artificial Pancreas: Current Algorithms and Homeostasis
No Thumbnail Available
Diabetes type 1 is a homeostatic imbalance caused by lack of insulin producing beta-cells in the pancreas. Standard treatment is very invasive, time-consuming, and error prone. One treatment that has been researched more and more is the artificial pancreas, or also called, a closed-loop system. Such a system results from the combination of a continuous glucose monitor, an insulin pump, and a control algorithm. Current state-of-the-art algorithms are still struggling with dealing with uncertainties, such as unannounced meals and exercise. For tighter blood glucose control and prevention of severe high and low blood glucose levels better algorithms are needed. More specifically, algorithms that are individualized and adaptable. The aim of this thesis was to get an overview of the current artificial pancreas being researched and identify challenges. In addition, identifying novel approaches that can deal with the challenges. Approaches that make use of both hormones insulin and glucagon manage to get a better control and have a higher chance of being insensitive to unannounced meals. There are several promising data-driven methods from machine learning such as reinforcement learning. Future research should be focused on data-driven approaches that can model the uncertainties in glucose homeostasis.
Faculteit der Sociale Wetenschappen