An anticipatory approach to optimal experimental design for model discrimination

A Global Parallel Model Based Design of Experiments Method to Minimize Model Output Uncertainty

“…This approach is generally called model-based design of optimal experiments (MBDOE) or quantitative experiment design (Franceschini and Macchietto 2008a;Galvanin et al 2007;Kreutz and Timmer 2009;Pronzato 2008). These approaches are largely designed to reduce the uncertainty in the mathematical model parameters as implemented in Franceschini and Macchietto (2008a), Pronzato and Walter (1994), Beck and Woodbury (1998), Emery et al (2000), Cho et al (2003), Kutalik et al (2004), Joshi et al (2006), Rodriguez-Fernandez et al (2006), Gutenkunst et al (2007), Chu and Hahn (2008), Lillacci and Khammash (2010), Van Derlinden et al (2010) and/or discriminate between rival models as focused on in Box and Hill (1967), Pritchard and Bacon (1974), Ferraris et al (1984), Chen and Asprey (2003), Kremling et al (2004), Vatcheva et al (2006), Donckels et al (2009). Franceschini and Macchietto (2008a) put forth a comprehensive experiment design paradigm that included an initial stage of preliminary investigations to evaluate structural and practical identifiability before proceeding to sequential, parallel, or sequential-parallel experiment design for model discrimination and parameter refinement.…”

A kernel‐based method to determine optimal sampling times for the simultaneous estimation of the parameters of rival mathematical models

“…In this section, the experimental design concepts introduced in the previous section will be illustrated in a relatively simple case study, 3 where nine models are proposed to describe the kinetic behavior of the enzyme glucokinase (glk, EC: 2.7.1.2). This enzyme catalyzes the conversion of glucose (GLU) and ATP to glucose-6-phosphate (G6P) and ADP, which is the first reaction of the glycolysis pathway.…”

“…This is because the experimental design methods are model-based, and high model prediction uncertainties obviously hamper the efficacy and efficiency of the model discrimination procedure. [1][2][3][11][12][13][14][15] These model prediction uncertainties are determined by the quality of the available data, since low-quality data will result in poorly estimated parameters, which on their turn result in uncertain model predictions. The discrimination among several rival models may thus become more efficient and effective if this uncertainty could be reduced prior to the start of the model discrimination procedure.…”

An ideal point method for the design of compromise experiments to simultaneously estimate the parameters of rival mathematical models

“…In this respect, the importance of the uncertainty on the model predictions cannot be overstated. Indeed, when this uncertainty is too large, the expected differences in the model predictions may not occur after all, which undermines the efficacy and efficiency of the model discrimination procedure (Burke et al, 1996(Burke et al, , 1997Buzzi-Ferraris et al, 1984;Chen and Asprey, 2003;Donckels et al, 2009a;Kremling et al, 2004;Schwaab et al, 2006;Ternbach et al, 2005).…”

“…Because the latter is usually time-and money-consuming, carefully designing these experiments can significantly reduce the required experimental effort. To achieve model discrimination in a minimal number of experiments, experimental design methods described in literature can be used (Buzzi-Ferraris et al, 1984;Chen and Asprey, 2003;Donckels et al, 2009a;Hunter and Reiner, 1965;Vanrolleghem and Van Daele, 1994).…”