Accuracy of non-linear kinetic models for predicting ruminal fermentation of agro-industrial by-products

Gas production kinetics is considered as one of the key indicators for assessing the nutritional value of feeds; therefore, precise prediction of kinetic parameters can provide reliable estimates of the nutritive value of feedstuffs. The objective of this study was to assess the accuracy of various nonlinear models in predicting ruminal fermentation parameters using the in vitro gas production (IVGP) technique. The fermentation substrates used in this study were agro-industrial by-products, including sugar beet pulp, lemon pulp, tomato pomace, grape pomace, sesame meal, rapeseed meal, bakery waste, and saffron flower waste. Rumen fluid was collected from three adult ruminally-fistulated Mehraban rams, then filtered and buffered. Each of the feed samples (in ۳ replicates and ۳ separate runs) was incubated with buffered rumen fluid for ۱۴۴ hours. The gas production data were fitted to five nonlinear models, including Exponential (EXP), Gompertz (GOM), Logistic (LOG), Mitscherling (MCH), and Weibull (WEB). The goodness of fit of these models was evaluated using metrics such as mean square error (MSE), coefficient of determination (R۲), residual mean absolute deviation (RMAD), and mean percentage error (MPE). Additionally, the Akaike's information criterion (AIC), Bayesian information criterion (BIC), accuracy factor (AF), run test, and linear regression analysis were employed to assess the accuracy of the models. Based on MSE and R۲ statistics, the EXP model demonstrated the lowest accuracy (۳۷.۳۰ and ۰.۹۵۸, respectively), while the MCH (۸.۱۵ and ۰.۹۹۱, respectively) and WEB (۴.۰۲ and ۰.۹۹۶, respectively) models exhibited the highest accuracy (P<۰.۰۵). Additionally, the AIC, BIC, and AF statistics were lowest for the WEB and MCH models, and highest for the EXP model. The results of the run test and linear regression analysis corroborated these findings. Overall, these findings indicated that the WEB model was the most accurate among the evaluated models for predicting the rumen fermentation kinetics, offering a reliable estimate of the nutritional value of new feedstuffs, such as agro-industrial by-products.

کلیدواژه ها:

Exponential model ، Prediction ، gas production kinetics ، goodness of fit