Paper Info
Title
Model predictive control with event-triggered communication for an embedded artificial pancreas
Abstract
Embedded or wearable artificial pancreas (AP) systems must be capable of low energy operation to ameliorate battery depletion and reduce the need for frequent recharge cycles. One of the major contributors to energy drain of such embedded AP systems is the power utilized for wireless communication among the components of the AP: the continuous glucose monitor (CGM), control module, insulin pump, and mobile app for display. In this paper, an event-triggered communication (ETC) algorithm is proposed for reducing sensor-to-controller transmissions. Concretely, a CGM value is not transmitted to the controller if the current and predicted glucose trajectory of the patient resides in a safe zone in the glucose space. An observer-based model predictive control (MPC) algorithm is subsequently deployed to regulate glucose with aperiodic, event-triggered CGM transmissions. Simulations on ten in-silico patients using the UVA/Padova metabolic simulator reveals that the proposed algorithm is capable of reducing the sensor-controller transmissions by around 50%. In spite of statistically significant communication reduction, the MPC with ETC maintains glucose values (mean ± standard deviation) in the clinically accepted range of 70180 mg/dL for 70.3±7.9% of the time in closed-loop, whereas a standard MPC maintains glucose in the euglycemic range for a comparable 71.8±7.9% with large, unannounced meals and nocturnal hypoglycemia. Furthermore, the proposed formalism exhibits lower time than the standard MPC in the hypoglycemic (<;70 mg/dL) range. Thus, the proposed approach communicates efficaciously without compromising patient safety.
Year
DOI
Venue
2017
10.1109/CCTA.2017.8062517
2017 IEEE Conference on Control Technology and Applications (CCTA)
Keywords
Field
DocType
battery depletion,wireless communication,continuous glucose monitor,insulin pump,mobile app,event-triggered communication algorithm,sensor-to-controller transmissions,glucose trajectory,model predictive control algorithm,CGM transmissions,sensor-controller transmissions,UVA-Padova metabolic simulator,wearable artificial pancreas systems,nocturnal hypoglycemia
Artificial pancreas,Control theory,Wireless,Simulation,Computer science,Model predictive control,Insulin pump,Observer (quantum physics),Standard deviation,Trajectory
Conference
ISBN
Citations 
PageRank 
978-1-5090-2183-3
0
0.34
References 
Authors
10
4
Name
Order
Citations
PageRank
Ankush Chakrabarty16812.73
Stamatina Zavitsanou281.64
Francis J Doyle324445.10
Eyal Dassau4386.55