Reducing parameter uncertainty in kinetic models: A new strategy for experimental design

Mechanistic models can be used to simulate, optimize, and control bioprocesses. The accuracy and reliability of their predictions strongly depend on the uncertainty of model parameters. The quality of estimates is therefore a crucial feature of the model used. There are many methods of experimental design, often based on linear regression and simple statistical reasoning; however, there is a lack of methods dedicated to designing experiments that provide optimal data sets for kinetic model calibration. It appears that parameters-to-data sensitivity coefficients (PSCs) can be used to determine when dependent variables need to be measured to achieve a good model fit. Therefore, in this study, various methods for determining PSCs and evaluating their properties were assessed to propose a new experimental design procedure. The new method enables a reduction in the number of measurements and the uncertainty of estimated parameters. It can be used to reduce the time and costs of experiments.