Skip to main content

Individualized targeted glucose control to avoid hypoglycemia


Hyperglycemia and hypoglycemia have been linked to worse outcomes in critically ill patients. While there is controversy as to the optimal tightness of glucose control in critically ill patients, there is agreement that an upper limit to safe glucose levels exists and that avoiding hypoglycemic episodes should be prioritized. Our algorithm can assist clinicians in maintaining blood glucose ([Gbl]) within a desired target range while avoiding hypoglycemia.


Our model predictive control (MPC) algorithm uses insulin and glucose as control inputs and a linearized model of glucose-insulin-fatty acid interactions. To allow the controller model to learn from data, a moving horizon estimation (MHE) technique tailored the tissue sensitivity to insulin to individual responses. Patient data ([Gbl] measurements, insulin and nutritional infusion rates) were from the HIDENIC database at the University of Pittsburgh Medical Center. [Gbl] measurements, typically hourly, were interpolated to impute a measurement every 5 minutes. The model captured patient [Gbl] via nonlinear least squares by adjusting insulin sensitivity (SI) and endogenous glucose production (EGP0). The resulting virtual patient (VP) is used to evaluate the performance of the MPC-MHE algorithm.


MPC controller performance on one VP is shown in Figure 1. Across a population of 10 VPs, the average [Gbl] under MPC is 6.31 mmol/l, the average minimum is 4.62 mmol/l, the population individual minimum is 3.49 mmol/l and the average absolute average residual error is 0.83 mmol/l from a 5.6 mmol/l target. With standard intervention, the 10 VPs have an average [Gbl] of 9.32 mmol/l, an average minimum [Gbl] of 3.77 mmol/l, and a population minimum [Gbl] of 2.78 mmol/l. Algorithm performance deteriorates significantly if the imputed sampling time exceeds 30 minutes, underlining the importance of dynamic variations in insulin sensitivity in this population

Figure 1



The MPC-MHE algorithm achieves targeted glucose control in response to changing patient dynamics and multiple measured disturbances for a pilot population of 10 VPs. Furthermore, the MHE scheme updates patient parameters in real time in response to changing patient dynamics.

Author information



Corresponding author

Correspondence to SP Gawel.

Rights and permissions

This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Cite this article

Gawel, S., Clermont, G., Ho, T. et al. Individualized targeted glucose control to avoid hypoglycemia. Crit Care 17, P456 (2013).

Download citation


  • Hypoglycemia
  • Model Predictive Control
  • Average Minimum
  • Hypoglycemic Episode
  • Virtual Patient