Processing weights of chickens determined by dual-energy X-ray absorptiometry: 3. Validation of prediction models

• Published in: Animal - Open Space 2022-12, Vol.1 (1), p.100022, Article 100022
• Main Authors: Martinez, D.A., Weil, J.T., Suesuttajit, N., Beitia, A., Maharjan, P., Hilton, K., Umberson, C., Scott, A., Coon, C.N.
• Format: Article
• Language: English
• Subjects: breast meat; breasts; dual-energy X-ray absorptiometry; equations; fasting; feed deprivation; fillets; meat production; prediction; ready-to-cook foods; regression analysis
• ISSN: 2772-6940; 2772-6940
• DOI: 10.1016/j.anopes.2022.100022

Dual-Energy X-ray Absorptiometry (DEXA) has been shown to predict the processing weights of the carcass and cut-up pieces of the chicken. This study aimed to validate models to predict processing weights using DEXA that were previously developed. An experiment was conducted with broilers grown up in 80 floor pens randomly subjected to one of five dietary treatments. On day 41, all birds were weighed, seven were randomly selected per pen, and their weights were recorded. After a feed withdrawal period, five selected birds per pen were transported to a processing plant, and the rest of the birds were fed again. The carcass was weighed before and after chilling for one hour. The chilled carcass was cut up, the weights of each commercial piece were recorded (breast fillet, tenders, wings, leg quarters, total white meat, and ready-to-cook parts), and the corresponding unchilled weights were calculated. On day 43 or 44, the other two birds selected per pen were weighed, DEXA-scanned without fasting, and their fasted weights were determined by applying a previously developed equation. Predicted processing weights were obtained by entering DEXA-reported values into a set of models previously developed. All the data were adjusted to the same BW basis. The pen mean observed processing weights and the DEXA-predicted ones were used to validate the models. The linear regression between predicted and observed values was calculated, and the R² was used as a precision index. The parallelism of the predicted and observed response curves across dietary treatments, and the model prediction error and accuracy were determined. The validation criteria were based on the validation R², the change in R² from development to validation, the parallelism of response curves, and the prediction accuracy. The validation R² of all tested models predicting the weight of cut-up pieces was > 0.84, and their prediction errors were ≤ 5.85 %, except for the model predicting the weight of the wings (prediction error > 10 %). All traits showed parallel trends when the response curves across treatments obtained with the DEXA-predicted values or the processing plant data were compared. In conclusion, all models but the one predicting the weight of wings satisfied the evaluation criteria and were validated, supporting the use of DEXA to determine the processing weights of broilers and its application to the study of nutrition interventions to improve breast meat production.