A computational model of the human glucose-insulin regulatory system

Keh-Dong Shiang; Fouad Kandeel

摘要：Objective:A computational model of insulin secretion and glucose metabolism for assisting the diagnosisof diabetes mellitus in clinical research is introduced. The proposed method for the estimation of parameters for a system of ordinary differential equations (ODEs) that represent the time course of plasma glucose and insulinconcentrations during glucose tolerance test (GTT) in physiological studies is presented. The aim of this studywas to explore how to interpret those laboratory glucose and insulin data as well as enhance the Ackerman math-ematical model. Methods: Parameters estimation for a system of ODEs was performed by minimizing the sumof squared residuals (SSR) function, which quantifies the difference between theoretical model predictions andGTT's experimental observations. Our proposed perturbation search and multiple-shooting methods were appliedduring the estimating process. Results: Based on the Ackerman's published data, we estimated the key param-eters by applying R-based iterative computer programs. As a result, the theoretically simulated curves perfectlymatched the experimental data points. Our model showed that the estimated parameters, computed frequency andperiod values, were proven a good indicator of diabetes. Conclusion: The present paper introduces a computa-tional algorithm to biomedical problems, particularly to endocrinology and metabolism fields, which involves twocoupled differential equations with four parameters describing the glucose-insulin regulatory system that Acker-man proposed earlier. The enhanced approach may provide clinicians in endocrinology and metabolism field in-sight into the transition nature of human metabolic mechanism from normal to impaired glucose tolerance.