Mathematical Modeling in Systems Medicine: New Paradigms for Glucose Control in Critical Care

Stress hyperglycemia occurs frequently in critical care patients and many of the harmful repercussions may be mitigated by maintaining glucose within a ``healthy'' zone. While the exact range of the zone varies, glucose below 80 $mg/dl$ or above 130 $mg/dl$ increases risk of mortality. Zone glucose control (ZGC) is accomplished primarily using insulin administration to reduce hyperglycemia. Alternatively, we propose also allowing glucose administration to be used to raise blood glucose and avoid hypoglycemia. While there have been attempts to create improved paradigms for treatment of stress hyperglycemia, inconsistencies in glycemic control protocols as well as variation in outcomes for different ICU subpopulations has contributed to the mixed success of glucose control in critical care and subsequent disagreement regarding treatment protocols. Therefore, a more accurate, personalized treatment that is tailored to an individual may significantly improve patient outcome. The most promising method to achieve better control using a personalized strategy is through the use of a model-based decision support system (DSS), wherein a mathematical patient model is coupled with a controller and user interface that provides for semi-automatic control under the supervision of a clinician. Much of the error and subsequent failure to control blood glucose comes from the failure to resolve inter- and intrapatient variations in glucose dynamics following insulin administration. The observed variation arises from the many biologically pathways that affect insulin signaling for patients in the ICU. Mathematical modeling of the biological pathways of stress hyperglycemia can improve understanding and treatment. Trauma and infection lead to the development of systemic insulin resistance and elevated blood glucose levels associated with stress hyperglycemia. We develop mathematical models of the biological signaling pathways driving fluctuations in insulin sensitivity and resistance. Key metabolic mediators from the inflammatory response and counterregulatory response are mathematically represented acting on insulin-mediated effects causing increases or decreases in blood glucose concentration. Data from published human studies are used to calibrate a composite model of glucose and insulin dynamics augmented with biomarkers relevant to critical care. The resulting mathematical description of the underlying mechanisms of insulin resistance could be used in a model-based decision support system to estimate patient-specific metabolic status and provide more accurate insulin treatment and glucose control for critical care patients