The Impact of a Responsive Endogenous Pancreas in Critical Care Glucose Control

Critical care patients suffer from stress hyperglycemia, but differ from ambulatory diabetics by the presence of a working, albeit malfunctioning, endogenous pancreas. Control systems for diabetics can be designed without concern for a pancreas, as the pancreas is no longer responsive; critical care glucose control algorithms must be able to respond to the working, and presumably healing, pancreas in the recovering patient. We construct herein a PID-derived algorithm to capture the pancreatic response of ICU patients. Using clinical data from UPMC, the PID parameters from intensive care unit (ICU) patients are shown to be both variable, and potentially malfunctioning, e.g., possessing an integral gain of 0 – clearly demonstrative of the loss of glucose control commonly observed in post-surgical ICU patients. The impacts of this endogenous responsivity, in the context of commonly employed diabetic challenges as administered to in silico virtual patients developed from clinical data, demonstrate the need to incorporate this responsive system in critical care controller design, and to design ICU controllers with adaptive elements to manage the return to health of critical care patients.