Neonatal Glycemic Control - Model Validation and In Silico Virtual Patient Trials (2009)
Type of ContentConference Contributions - Other
PublisherUniversity of Canterbury. Mathematics and Statistics
University of Canterbury. Mechanical Engineering
AuthorsLe Compte, A.J., Lynn, A., Chase, J.G., Shaw, G.M., Russel, G., Blakemore, A., Lee, D.S., Wong, X.W.show all
Introduction: Premature, low-birth-weight infants in the neonatal intensive care unit (NICU) can lose blood glucose homeostasis due to immaturity of endogenous regulatory systems and the stress of their condition. Typical treatment relies on glucose restriction before insulin administration due to fear of hypoglycaemia. A model of the fundamental glucose regulatory dynamics could enable optimised treatment approaches.
Methods: An adult critical care metabolic system model is adapted to the unique physiological case of the neonate. Integral-based methods identify time-varying insulin sensitivity and non-insulin mediated glucose uptake profiles for virtual patient trials. Retrospective clinical for N=25 cases contained 1079 glucose measurements over 3589 total patient hours plus all insulin and nutritional infusion data. Birth weights were all less than 1.5 kg and gestational age was 23-28.6 weeks. The model is validated for predictions of 1-4 hours forward. Virtual patients are used to develop model-based glycaemic control protocols.
Results: The identified model had a median absolute percentage error of 2.50% [IQR: 1.0%-5.3%]. Median absolute prediction errors at 1, 2 and 4-hour intervals were 5.8% [IQR: 2.6%-11.2%], 9.9% [IQR: 4.5%-19.3%] and 14.5% [IQR: 6.3%-27.2%] respectively. Virtual trial results targeting a 72-125 mg/dL range yield a median blood glucose level of 104 mg/dL [IQR: 90-117]. Average insulin usage is 0.069 U/kg/hr. These results are compared to clinical values of 144 mg/dL [IQR: 113-178] mg/dL and 0.034U/kg/hr. Average dextrose delivery was 8.9 mg/kg/min. Hypoglycemic events were minimised in virtual trials.
Conclusions: The model accurately captures and predicts the fundamental dynamic behaviours of the neonatal metabolism well enough for effective potential clinical use in glycaemic control. Model-based control can offer improved control with greater nutrition delivery to provide potentially better long-term outcomes.
CitationLe Compte, A.J., Lynn, A., Chase, J.G., Shaw, G.M., Russel, G., Blakemore, A., Lee, D.S., Wong, X.W. (2008) Neonatal Glycemic Control - Model Validation and In Silico Virtual Patient Trials. Bethesda, MD, USA: 8th Annual Diabetes Technology Meeting, 13-15 Nov 2008.
This citation is automatically generated and may be unreliable. Use as a guide only.