Processing weights of chickens determined by dual-energy X-ray absorptiometry: 2. Developing prediction models

A considerable opportunity exists in evaluating the dynamics of the carcass and the processing cut-up weights of broilers across the whole grow-out period as influenced by intervention factors. However, no fast and objective tool exists up to date to make such determinations. This study aimed to develop models to predict the unchilled and chilled weights of the carcass and cut-up pieces of broilers using Dual-Energy X-ray Absorptiometry (DEXA) and feathered non-fasted birds. Highly diverse (BW and body composition) broilers (n = 291) between 4 and 79 days of age were euthanized, DEXA-scanned, and manually processed to determine the weights of the carcass and cut-up pieces. Correction factors were applied to obtain the fasted BW and the corresponding bled and chilled weights. A database was built up, including all the weights recorded and the DEXA-reported indexes. A stratified random data-splitting with a refitting approach was applied. Multiple least-squares linear regressions were fitted for each unchilled and chilled variable on the training dataset using JMP Pro 16. Natural log and square root transformations were applied to predictor variables as convenient, and outliers were removed. Candidate models were screened for normal distribution and homoscedasticity of residuals and collinearity among predictors. The highest precision (adjusted R²) and the lowest error (RMSE) were selection criteria. Once model overfitting and prediction performance was tested on the validation dataset, the models were refitted with all the data in the original dataset. Prediction models with high (unchilled and chilled carcass and cut-up weights, feet, and head; R² > 0.99) and acceptable (abdominal fat; R² > 0.69) precision were obtained. In conclusion, these results support the use of DEXA to determine the processing weights of broilers. Its application to the study of growth curves of cut-up pieces as influenced by nutrition, genetics, environment, and management opens a new spectrum of opportunities for the industry.