A novel optimal experiment design technique based on multi-objective optimization and its application for toxin kinetics model of hemodialysis patients

Abstract

It is increasingly recognized that rapid model development or the refining of currently available models for any system (chemical, biological, medical, environmental etc.) is greatly facilitated by experimental design and particularly by Model Based Optimal Experiment Design (MBOED). MBOED approaches are primarily used in the context of precise parameter estimation and model discrimination. However, the presence of parameter correlation often reduces the confidence in creating practical applications based on the identified model. To overcome this problem, a novel design criterion comprising two conflicting objectives: (1) maximizing the information content (which is the usual objective in traditional OED), and (2) minimizing the correlation between estimated parameters, is proposed. The proposed approach is implemented to suggest the optimal sampling times for estimating the model parameters of toxin kinetics model for patients on maintenance hemodialysis, developed recently by Maheshwari et al., 2011. The resulting parameter estimates can be used for prescribing patient specific dialysis sessions. The proposed multi-objective MBOED criterion is also relevant for use in other biomedical, chemical, and environmental systems. © 2012 Elsevier B.V.