Overnight Control of Blood Glucose in People with Type 1 Diabetes

Abstract

Abstract:In this paper, we develop and test a Model Predictive Controller (MPC) for overnight stabilization of blood glucose in people with type 1 diabetes. The controller uses glucose measurements from a continuous glucose monitor (CGM) and its decisions are implemented by a continuous subcutaneous insulin infusion (CSII) pump. Based on a priori patient information, we propose a systematic method for computation of the model parameters in the MPC. Safety layers improve the controller robustness and reduce the risk of hypoglycemia. The controller is evaluated in silico on a cohort of 100 randomly generated patients with a representative intersubject variability. This cohort is simulated overnight with realistic variations in the insulin sensitivities and needs. Finally, we provide results for the first tests of this controller in a real clinic.

Tuning of Controller for Type 1 Diabetes Treatment with Stochastic Differential Equations

“…meals and exercise are minimal at this time of day. In [Boiroux et al (2012)] details about the setup is described. The subject wore an insulin pump (Medtronic Paradigm Veo, Minneapolis, USA) and two CGMs (Dexcom Seven Plus, San Diego, USA) measuring the glucose level every five minutes.…”

Comparison of Prediction Models for a Dual-Hormone Artificial Pancreas∗∗Funded by the Danish Diabetes Academy supported by the Novo Nordisk Foundation. Contact information: John Bagterp J_rgensen(jbjo@dtu.dk).

“…For instance, Kirchsteiger et al (2011) used a third order transfer function with an integrator, van Heusden et al (2012) used a third order discrete transfer function model and Percival et al (2010) applied a first order transfer function with a time delay and an integrator. In our previous work, we used a second order transfer function model, see Boiroux et al (2012); Bátora et al (2015). All these models have been validated on simulations and/or clinical studies.…”

Model-Based Closed-Loop Glucose Control in Type 1 Diabetes: The DiaCon Experience

“…Details about the controller have been reported elsewhere. 18,19 The deterministic part of the algorithm is specified by three empirically determined patient-specific parameters: insulin sensitivity factor (ISF), insulin action time, and basal insulin infusion rate at the time of CL initiation. Estimation of the stochastic part of the model was based on data from one diabetes patient.…”

“…Each study group has its own strategy for CL control: different glucose sensors and insulin-delivery devices are in use; different mathematical methodologies are applied in the construction of control algorithms; different platforms communicating data between system components have been built; and some groups have extended their system with even further components such as glucagon delivery. [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] Impressive work has been conducted by these groups, and landmark results have been achieved, yet there is still a distance to cover before a safe, robust, and fully automated CL system becomes a reality.…”

“…We developed an algorithm based on model predictive control (MPC) theory and conducted the first clinical feasibility study in 2011-2012. 18,19 This article presents our approach to CL glucose control, including a description of the clinical setup, details of the control algorithm, and results from the first clinical test of the system. In addition, we discuss differences in CL study protocols and procedures, which pose difficulties with regard to control algorithm performance comparisons.…”