Embedded Control in Wearable Medical Devices: Application to the Artificial Pancreas

Significant increases in processing power, coupled with the miniaturization of processing units operating at low power levels, has motivated the embedding of modern control systems into medical devices. The design of such embedded decision-making strategies for medical applications is driven by mult...

Full description

Saved in:
Bibliographic Details
Published in:Processes 2016-12, Vol.4 (4), p.35
Main Authors: Zavitsanou, Stamatina, Chakrabarty, Ankush, Dassau, Eyal, Doyle, Francis
Format: Article
Language:English
Subjects:
Comparative studies
Computer applications
Control systems
Controllers
Decision making
Diabetes mellitus
Diabetes mellitus (insulin dependent)
Embedded systems
Embedding
Glucose
Maintenance costs
Medical devices
Medical equipment
Miniaturization
Pancreas
Power consumption
Predictive control
Quadratic programming
Safety critical
System failures
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Significant increases in processing power, coupled with the miniaturization of processing units operating at low power levels, has motivated the embedding of modern control systems into medical devices. The design of such embedded decision-making strategies for medical applications is driven by multiple crucial factors, such as: (i) guaranteed safety in the presence of exogenous disturbances and unexpected system failures; (ii) constraints on computing resources; (iii) portability and longevity in terms of size and power consumption; and (iv) constraints on manufacturing and maintenance costs. Embedded control systems are especially compelling in the context of modern artificial pancreas systems (AP) used in glucose regulation for patients with type 1 diabetes mellitus (T1DM). Herein, a review of potential embedded control strategies that can be leveraged in a fully-automated and portable AP is presented. Amongst competing controllers, emphasis is provided on model predictive control (MPC), since it has been established as a very promising control strategy for glucose regulation using the AP. Challenges involved in the design, implementation and validation of safety-critical embedded model predictive controllers for the AP application are discussed in detail. Additionally, the computational expenditure inherent to MPC strategies is investigated, and a comparative study of runtime performances and storage requirements among modern quadratic programming solvers is reported for a desktop environment and a prototype hardware platform.
ISSN:2227-9717
2227-9717
DOI:10.3390/pr4040035