Investigation of glucose-insulin model efficacy for diabetes patient in the ICU

Abstract

This paper investigates the inter-relationship of blood glucose (BG) model fitting error percentage for both diabetic and non-diabetic patient’s status with ICING glucose-insulin model efficacy. Data were taken from 132 critically ill patients admitted in ICU of Hospital Tunku Ampuan Afzan (HTAA), Kuantan, Malaysia. The BG fitting error results are important to determine the efficacy and safety of chosen glucose-insulin model-based. Analysis results of per-patient and by-cohort were used to suggest whether diabetes is an influential factor for model BG fitting error efficacy. For diabetic per-patient and by-cohort, median model fit error are 0.29 and 0.31% respectively Meanwhile, for non-diabetic per-patient and by-cohort model fit error are 0.27 and 0.29% respectively. In conclusion, results demonstrated no significant difference between diabetic and non-diabetic patient in model fitting error. Thus, ICING model may be used to deliver real time model-based insulin therapy whether patient is diabetic or non-diabetic. © 2018, Springer Science+Business Media Singapore.