Determining Optimal Nonlinear Regression Models for Studying the Kinetics of Fatty Acid Ruminal Biohydrogenation In Vitro

The accurate estimation of in vitro ruminal biohydrogenation (BH) kinetics of fatty acids (FA) allows for a more accurate understanding of their dynamics and develop targeted strategies to enhance desirable FA bypass. This study comprises a comprehensive evaluation of 33 nonlinear regression models to determine the most suitable model for accurately estimating the in vitro BH kinetics of individual FA. The data set utilized in the present research originates from a recent investigation on the effects of micronization and vitamin E on the in vitro ruminal BH of rapeseed. For the nonlinear regression analysis, data comprising FA concentrations (expressed as g FA/100 g FA) at the conclusion of 2, 4, 8, 12, 24, and 48 h incubation periods were employed. The evaluation of nonlinear regression models focused on identifying the ideal model based on criteria including the highest R 2 value, the lowest RMSE value, and statistically significant coefficients. The results pinpoint the Gompertz model as an effective choice for estimating the in vitro ruminal BH kinetics of upward-trending fatty acids, including intermediate unsaturated fatty acids and saturated end FA. Additionally, the first-order kinetic model of Ørskov and McDonald emerges as the preferred model for investigating the BH kinetics of downward-trending fatty acids, including oleic acid, linoleic acid, and alpha-linolenic acid. In summary, this rigorous evaluation led to the identification of the most appropriate model, one that not only exhibited an exceptional fit to the data but also provided profound insights into the intricate relationships between predictors and the dynamic behavior of FA. The established nonlinear regression models will serve as invaluable tools for future research investigating FA biohydrogenation kinetics.