Prediction of metabolizable energy content of poultry feedstuffs – response surface methodology vs. Artificial neural network approach

Abstract

Metabolisable energy (ME) represents portion of energy utilized by the animal. Experiments for determination of ME require test animals, collection of samples and excreta, and determination of total energy content of used material. Therefore, ME determination can be expensive and time consuming. The aim of this study was to investigate the effect of enzymatic digestible organic matter (EDOM) and values of proximate chemical analysis on prediction of true metabolisable energy (TME) of feedstuffs for broilers. The performance of Artificial Neural Networks (ANN) was compared with the performance of second order polynomial (SOP) model, as well as with experimental data in order to develop rapid and accurate method for prediction of TME. Analysis of variance and post-hoc Tukey’s HSD test at 95% confidence limit have been calculated to show significant differences between different samples. Response Surface Method has been applied for evaluation of TME. Second order polynomial model showed high coefficients of determination (r2 = 0.927). ANN model also showed high prediction accuracy (r2 = 0.983). Principal Component Analysis was successfully used in prediction of TME.