Kinetic modelling in food science : a case study on chlorophyll degradation in olives

This paper critically reviews the use of kinetics in modelling (bio)chemical changes in foods. Multiresponse modelling (taking more reactants and products of the reaction under study into consideration in the modelling process) is advocated as opposed to uniresponse modelling (in which only one response, a reactant or a product, is considered). Multiresponse modelling requires slightly more sophisticated software but is rewarding in that models can be tested much more rigorously and that estimation of parameter values is much more precise than can be achieved with uniresponse modelling. The approach was applied to an example taken from the literature, degradation of chlorophyll in olives during fermentation. The two models proposed in the literature for this degradation were found to be incorrect using the multiresponse approach. An alternative model was proposed which gave a much better fit. In addition, ways were indicated how the model could be further refined. It was also shown with this example that transformation of data (such as taking the logarithm to apply a first-order model) can lead to false conclusions concerning the kinetic model and consequently to serious errors in the resulting parameter estimates. In almost all cases, non-linear regression with untransformed data is to be preferred.