Addressing the Identifiability of Parameters in the Michaelis-Menten Model through Bayesian Optimal Design
Enzyme-mediated biochemical reactions are fundamental in various biological processes. The Michaelis-Menten model is a representation of enzymatic reaction rates concerning substrate concentration, which provides information on these reactions. However, the estimation of its parameters is affected by identifiability issues, which relate to a high correlation between the estimators and/or with the differential variability of them. Different models have been explored to mitigate these issues, but such efforts have been insufficient. This work proposes the optimal experiment design approach, with the objective of improving both the sphericity of the confidence ellipsoid as well as reducing the correlation of parameter estimators, through a comparative analysis between D-optimal and R-optimal criteria under Bayesian approaches. This comparison is made using correlation indicators of the estimators (frequentist and Bayesian) and variance ratio between the error estimates.
Keywords: Progress Curve Assay Identifiability Problems Correlation of parameters Nonlinear Optimization.