Comparing different methods for estimating kinetic parameters of whey protein heat-induced denaturation in infant milk formulas

摘要

Modeling the heat-induced denaturation of milk proteins is a relevant issue because of heating processes in the manufacturing of several dairy products. In this study, four different parameter estimation methods were evaluated to estimate the kinetic parameters of the heat-induced protein denaturation of beta-lactoglobulin (beta-LG) and lactoferrin (LF) in infant milk formulas (IMF). The methods were: a two-step method, nonlinear least-squares (NLS), one-step linearized, and weighted least-squares (WLS). The WLS was the best alternative, avoiding biases observed when applying other methods besides producing consistently low residuals. Using the second proposed weight function, the WLS produced sum of squared errors and mean absolute percentage errors with average values of respectively 0.18 and 12.3% versus 0.27 and 13.3% considering all methods. However, the WLS requires good initial guesses for parameters, and previous knowledge of sampling and residuals variance. Those can be provided by previously performing the two-step and NLS methods, respectively.