Fault-tolerant Control Systems for Artificial Pancreas

Principal Investigators

Dates: 9/30/13 – 6/30/18

Co-Investigators: Laurie Quinn, Joan Briller, Lisa Sharp

Abstract: Recent progress in artificial pancreas (AP) development produced various AP system concepts that can automatically regulate the blood glucose concentrations (BGC) of patients with Type 1 Diabetes Mellitus (T1DM).  The performance of these AP systems in clinical studies has been very encouraging.  These studies have also indicated the need for improving the reliability of APs in routine use by patients during their daily activities.  Large-scale adoption of AP systems by patients with T1DM requires the availability of reliable AP systems that can function over long periods of time without putting the user at risk of hyper- or hypoglycemia caused by system malfunctions. The hypothesis of our research is that reliable AP systems can be designed by integrating powerful fault detection and diagnosis, physical and analytical redundancy, and fault-tolerant control techniques to provide self-recovery, safeguards against failures and warnings and messages to users and care providers.

An AP includes various components such as sensors, continuous glucose monitors, pumps and infusion sets that are prone to failures. It also has software for interpreting sensor information, calculating the insulin on board, and computing the dose of insulin to be infused by the pump. The development of algorithm-based approaches to detect, diagnose and mitigate failure events in the operation, components or software of AP systems is necessary to maintain euglycemia in spite of failure(s) in hardware, software, mechanics or electronics of an AP.  The ultimate goal is to develop a fault-tolerant AP that can mitigate the risks in AP systems and function at an acceptable level for BGC regulation until the diagnosed fault can be repaired or that can provide safe transfer to manual operation of insulin pumps by the user. We will collaborate with industry to integrate our methods and tools with their hardware and software for translation to the ultimate development of a fault-tolerant AP system.

The focus of the proposed research is on the development of algorithms and software tools for performance monitoring, fault detection and diagnosis, control system performance assessment, analytical redundancy in sensors, control algorithms with fault-tolerance and recovery, and warning systems to users and care providers to a create fault-tolerant AP system.  The critical characteristics of our approach are (1) the use of physiological variable information to complement BGC, (2) multivariable modeling, monitoring, diagnosis and control techniques, and (3) recursive models and adaptive model-based control systems.  The proposed research will be a collaboration between Ali Cinar – Illinois Institute of Technology, Elizabeth Littlejohn – University of Chicago, and Laurie Quinn – University of Illinois at Chicago.

Hypothesis:  Safe and reliable AP systems can be designed by developing a fault-tolerant AP system with performance monitoring, fault detection and diagnosis, physical and analytical redundancy, and fault-tolerant control modules that coordinate their activities to sense abnormal situations, automatically find solutions to address the problem, generate warnings to users and medical personnel and if necessary provide safe transfer to manual operation of insulin pumps by the user.

The specific aims of the proposed research are:

Aim 1: Development of a performance monitoring system to detect abnormal situations in the BGC of the users and in the operation of the AP system.

Aim 2: Development of fault detection and diagnosis system to identify a fault and its source cause in the operation of the AP system.  The following will be developed and integrated to achieve this aim:

  • A controller performance assessment system,
  • Fault detection and fault diagnosis systems for sensors and the insulin pump,
  • Fault detection and diagnosis systems for sensor placement and insulin infusion to the body
  • Fault detection and diagnosis systems in signal transmissions
Illinois Institute of Technology