Prediction of nitrogen-corrected true metabolizable energy based on chemical composition in various wheat bran samples using multiple linear regression

The purpose of this study was to develop multiple linear regression (MLR) model to predict the nitrogen-corrected true metabolizable energy (TMEn) value of wheat bran. The amount of crude fat, ash, crude protein, crude fiber (all used as % of DM) and TMEn (Kcal/kg DM) were measured in 25 wheat bran samples with 4 replicates. The force-fed method has been used to estimate TMEn and excreta were collected for 48 h. There were significant (P < 0.001) differences in chemical composition and TMEn of wheat bran samples. The average crude fat, ash, crude protein, crude fiber and TMEn content of samples was determined to be 4.80, 5.68, 16.23, 8.60 (all used as % of DM) and 2062 (Kcal/kg DM), respectively. The calculated MLR model to predict the TMEn value (Kcal/kg) based on chemical composition (% of DM) was obtained as follows: TMEn = 2364 + (19×crude protein) + (46.1×crude fat) – (63×crude fiber) – (51.1×ash). The R2 value revealed that developed model could accurately predict the TMEn of wheat bran samples (R2=0.82). Crude fat and crude protein had a positive effect on TMEn, while ash and crude fiber had a negative impact on TMEn. The sensitivity analysis on the model indicated that dietary crude fiber (%) is the most important variable in the TMEn, followed by dietary ash, crude fat and crude protein. The results suggest that the MLR model may be used to accurately estimate the TMEn value of wheat bran from its corresponding chemical composition.