Integrating Compartment Models with Recursive System Identification

Abstract

In this work, an adaptive and personalized compartment model that translates the abrupt bolus and discrete basal changes into estimates of plasma insulin concentration (PIC) is integrated with a recursive subspace-based system identification approach to characterize the transient dynamics of glycemic measurements. The PIC estimates are obtained through a compartment model with time-varying parameters and obscure carbohydrate consumption effects simultaneously estimated using a filtering algorithm. The estimated PIC is subsequently employed in the recursive system identification approach to determine a linear, time-varying model for predicting glucose measurements. Simulation case studies demonstrate the improvement in prediction ability of the proposed approach (mean absolute error [MAE] of 9.29 mg/dL) compared to the standard method (MAE of 10.19 mg/dL).