Estimation and modeling true metabolizable energy of sorghum grain for poultry

Abstract

Sorghum grain is an important ingredient in poultry diets. The TMEn content of sorghum grain is a measure of its quality. As for the other feed ingredients, the biological procedure used to determine the TMEn value of sorghum grain is costly and time consuming. Therefore, it is necessary to find an alternative method to accurately estimate the TMEn content. In this study, 2 methods of regression and artificial neural network (ANN) were developed to describe the TMEn value of sorghum grain based on chemical composition of ash, crude fiber, CP, ether extract, and total phenols. A total of 144 sorghum samples were used to determine chemical composition and TMEn content using chemical analyses and bioassay technique, respectively. The values were consequently subjected to regression and ANN analysis. The fitness of the models was tested using R2 values, MS error, and bias. The developed regression and ANN models could accurately predict the TMEn of sorghum samples from their chemical composition. The goodness of fit in terms of R2 values corresponding to testing and training of the ANN model showed a higher accuracy of prediction than the equation established by regression method. In terms of MS error, the ANN model showed lower residuals distribution than the regression model. The results suggest that the ANN model may be used to accurately estimate the TMEn value of sorghum grain from its corresponding chemical composition.