Open-Source Artificial Pancreas Testbed (available)

Starting Date: June 2019
Duration: 12 weeks
Time commitment: Full time
Prerequisites: Desirable: programming experience with Android or Node.js

The Artificial Pancreas (AP) is an automated system for delivering insulin therapy in Type 1 Diabetes (T1D) patients (whose pancreas cannot produce insulin on its own) [1]. The system comprises an infusion pump for insulin release, a so-called continuous glucose monitor (CGM) that senses glucose levels underneath the skin, and a control algorithm (running on the pump or an external device) that, based on the glucose readings, computes the optimal amount of insulin to deliver. Thus, the CGM sensor and the insulin pump are connected in closed-loop by the control algorithm.

In the last five years, a number of patient-driven open-source projects were launched with the aim to make the AP technology widely available to the T1D community (#WeAreNotWaiting) [2]. These projects focus on developing AP control algorithms and interfacing with existing commercial pumps and CGMs. The main open-source DYI AP systems are OpenAPS (running on Raspberry Pi and other single-board computers) [3], AndroidAPS (Android systems) [4], and Loop (iOS) [5]. It was thanks to the success of these projects that device manufacturers have also begun developing closed-loop AP systems.

The goal of the project is to develop a testbed based on either OpenAPS or AndroidAPS for simulating AP control algorithms. After a conducting a preliminary review of current AP systems and a comparison of these two open-source platforms, you will choose the platform that can be best adapted to work with simulation models instead of the real hardware. You will install and configure the chosen AP platform, develop simple programs that simulate the CGM and the pump, and extend the AP platform to communicate with these programs. If time permits, you will incorporate more advanced control and meal detection algorithms from existing literature [6,7].

[1] Thabit, Hood, and Roman Hovorka. “Coming of age: the artificial pancreas for type 1 diabetes.” Diabetologia 59.9 (2016): 1795-1805. Available at
[2] Open Artificial Pancreas System | Dana Lewis | TEDxFHKufstein
[3] – #WeAreNotWaiting to reduce the burden of Type 1 diabetes.
[4] AndroidAPS — AndroidAPS 2.0.0 documentation.
[5] LoopDocs.
[6] Paoletti, Nicola, et al. “Data-driven robust control for type 1 diabetes under meal and exercise uncertainties.” International Conference on Computational Methods in Systems Biology. Springer, Cham, 2017.
[7] Chen, Hongkai, et al. “Committed Moving Horizon Estimation for Meal Detection and Estimation in Type 1 Diabetes.” To appear in Americal Control Conference 2019.