Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Model-Based Closed-Loop Glucose Control in Critical Illness

Knab, Timothy D (2017) Model-Based Closed-Loop Glucose Control in Critical Illness. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Download (10MB) | Preview


Stress hyperglycemia is a common complication in critically ill patients and is associated with increased mortality and morbidity. Tight glucose control (TGC) has shown promise in reducing mean glucose levels in critically ill patients and may mitigate the harmful repercussions of stress hyperglycemia. Despite the promise of TGC, care must be taken to avoid hypoglycemia, which has been implicated in the failure of some previous clinical attempts at TGC using intensive insulin therapies. In fact, a single hypoglycemic event has been shown to result in worsened patient outcomes.

The nature of tight glucose regulation lends itself to automatic monitoring and control, thereby reducing the burden on clinical staff. A blood glucose target range of 110-130 mg/dL has been identified in the High-Density Intensive Care (HIDENIC) database at the University of Pittsburgh Medical Center (UPMC). A control framework comprised of a zone model predictive controller (zMPC) with moving horizon estimation (MHE) is proposed to maintain euglycemia in critically ill patients. Using continuous glucose monitoring (CGM) the proposed control scheme calculates optimized insulin and glucose infusion to maintain blood glucose concentrations within the target zone.

Results from an observational study employing continuous glucose monitors at UPMC are used to reconstruct blood glucose from noisy CGM data, identify a model of CGM error in critically ill patients, and develop an in silico virtual patient cohort. The virtual patient cohort recapitulates expected physiologic trends with respect to insulin sensitivity and glycemic variability. Furthermore, a mechanism is introduced utilizing proportional-integral-derivative (PID) to modulate basal pancreatic insulin secretion rates in virtual patients. The result is virtual patients who behave realistically in simulated oral glucose tolerance tests and insulin tolerance tests and match clinically observed responses.

Finally, in silico trials are used to simulate clinical conditions and test the developed control system under realistic conditions. Under normal conditions the control system is able to tightly control glucose concentrations within the target zone while avoiding hypoglycemia. To safely counteract the effect of faulty CGMs a system to detect sensor error and request CGM recalibration is introduced. Simulated in silico tests of this system results in accurate detection of excessive error leading to higher quality control and hypoglycemia reduction.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Knab, Timothy Dtdk17@pitt.edutdk17
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairParker, Robert
Committee CoChairClermont,
Committee MemberMcCarthy, Joseph
Committee MemberCole,
Date: 26 September 2017
Date Type: Publication
Defense Date: 13 February 2017
Approval Date: 26 September 2017
Submission Date: 24 July 2017
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 275
Institution: University of Pittsburgh
Schools and Programs: Swanson School of Engineering > Chemical Engineering
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Glucose Control, Model Predictive Control, Critical Care, Mathematical Modeling
Date Deposited: 26 Sep 2017 19:32
Last Modified: 14 Mar 2023 15:46


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item