Software for in Silico Testing of an Artificial Pancreas



AbstractTight glycemic control of people suffering from Type 1 diabetes remains a challenge despite recent developments within diabetes treatment. A fully automated closed-loop Artificial Pancreas (AP) has the potential to become the new standard of care for people with type 1 diabetes. This through a tight glycemic control which resembles the control of a healthy individual. The road towards a commercial-viable AP is long and costly but In silico simulations has the potential to speed up the development of a commercial-viable AP.
The purpose of this thesis is to develop a software and a simulation framework which can become a benchmarking platform and an experimental playground for future in silico testing of the AP. This ensures a consistent testing standard of the AP, while mimicking the setup of clinical trials at a fraction of the cost. The simulation framework developed consist of 6 principal components;
1. A cohort of virtual patients (n = 50 adults) based on actual patient data
spanning the observed metabolic variability of key physiological parameters.
2. A mathematical model to describe the glucose-insulin dynamics of a patient
suffering from Type 1 Diabetes.
3. A model representing the usage of a Continuous Glucose Monitor.
4. A model to represent the usage of a subcutaneous insulin pump.
5. The implementation of an insulin administration strategy.
6. A set of outcome metrics allowing investigation of the performance of the AP.
These 6 principal components provides a comprehensive in silico simulation framework for closed-loop testing. Thus, a time- and cost-eective tool for performance testing of various APs tested in various user-defined scenarios prior to their clinical implementation.
The software has been implemented in the programming language Java using the open source numerical library Apache Commons Mathematics. The software has been constructed such that execution from Matlab is possible. Additionally, the software has been optimised to ensure fast runtime by the implementation of multi-threading.
The simulation framework would need future clinical validation. However, the findings from this thesis show that realistic computer simulations can provide valuable information about safety and the limitation on closed-loop control systems in a time- and cost-eective way. Hereby, the simulation software developed in this thesis serves as a strong tool in guiding and assessing clinical studies, why this thesis brings an ecient testing tool to the development of the AP.
TypeMaster's thesis [Academic thesis]
Year2014
PublisherTechnical University of Denmark, Department of Applied Mathematics and Computer Science
AddressRichard Petersens Plads, Building 324, DK-2800 Kgs. Lyngby, Denmark, compute@compute.dtu.dk
SeriesDTU Compute M.Sc.-2014
Note
Electronic version(s)[pdf]
Publication linkhttp://www.compute.dtu.dk/English.aspx
BibTeX data [bibtex]
IMM Group(s)Scientific Computing