Glycemic Control for Critically Ill Patients Using Zone Model Predictive Control

Hyperglycemia in critically ill patients leads to severe complications and even death. It was shown that keeping blood glucose (BG) within the range of 80–110 mg/dl (4.4–6.1 mmol/l) reduces mortality and morbidity in intensive care units (ICUs), and many researches for achieving such BG control have been performed. However, maintaining BG within such range is not easy mainly due to the time variability of insulin sensitivity in ICU patients. In this study, regarding hypoglycemia prevention as important, we improve our glycemic control system utilizing zone model predictive control, which is a control strategy to keep the controlled variable within a target range and has been applied to avoid hypoglycemia for type 1 diabetes mellitus. Then, we assess the performance of the glycemic control system on ICU virtual patients created from clinical ICU data. The modified glycemic control system achieves 68% of BG measurements within the range of 80–110 mg/dl and only 1.3% of BG measurements below 80 mg/dl, which suggests the effectiveness and safety of the system. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.