Using Modeling and Simulation in the Design of Closed-Loop Insulin Delivery System

Abstract

The complexity of medical devices is rapidly increasing, enabled by innovations in technology. This same technology has enabled health care to expand from institutional environments to home care and mobile environments. Software plays a critical role in controlling these devices. It is essential that the dependability of these devices is assured because many of them are safety critical; i.e. can cause serious injury or death. Modeling and simulation of device designs is emerging among medical device manufacturers as a technique to help address such challenges. Modeling and simulation has a long history of improving product quality by helping designers to detect defects that may be overlooked in a traditional software development workflow. For example, design requirements can be modeled, and then the model checked for properties of requirement consistency and completeness. Subsequently, the model can be run in-silica permitting rapid-prototyping and comprehensive real-time checking of design behavior which can ultimately be used to help verify actual device implementation. Modeling can be thought of as an incremental process improvement step that happens early in a design phase well before any implementation (writing code or building physical hardware), and can be integrated into any software/hardware development lifecycle. Medical Devices manufacturers and companies in other safety-critical industries such as aerospace and automotive have been using modeling and simulation approaches to verify and validate designs early in their development process for many years now [4][7][8]. Today, many diabetic patients use insulin pumps to manage their own medication delivery through a combination of basal and bolus doses of insulin to maintain healthy blood glucose (BG) levels. With the addition of Continuous Glucose Monitoring (CGM) devices, the opportunity exists to create a closed-loop, automatic control or “artificial pancreas” means of delivering the insulin. In this technical brief, we present a highly abstracted insulin infusion pump system as an example modeling exercise. A simulated environment of the glycemic system of the body, along with the insulin computation algorithm are modeled in Simulink® and tested against various real-world parameters from clinical testing.