Stage-specific prediction of body weight in Hanwoo steers using body measurements and nutrient intake variables

Accurate body weight (BW) prediction is important for growth management and nutrient utilization in Hanwoo steers. However, prediction performance may vary across growth stages and input variables. Therefore, this study evaluated stage-specific machine learning models for BW prediction using body measurements and nutrient intake variables.

Data from 136 Hanwoo steers were collected during the growing (6-12 months), early fattening (13-21 months), and late fattening (22-31 months) periods. Body size traits and nutrient intake variables were used to develop prediction models using Random Forest (RF), Linear Regression (LR), K-Nearest Neighbors (KNN), and Artificial Neural Networks (ANN). To minimize repeated-measurement bias, datasets were divided into training and testing sets based on individual animal identification. Prediction performance was evaluated using coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE).

BW showed strong associations with skeletal-related body traits during the growing period, whereas nutrient intake variables became more strongly associated with average daily gain (ADG) during fattening. Among the evaluated ML models, RF generally demonstrated robust predictive performance across growth stages. For BW prediction, RF achieved the highest accuracy during the growing period (R² = 0.952), whereas LR achieved the highest accuracy during the early-fattening period (R² = 0.947). ADG prediction performance was generally lower than that for BW. Variable importance analysis indicated that skeletal-related body traits contributed most to BW prediction, whereas nutrient intake variables contributed more to ADG prediction during fattening.

Factors associated with BW and ADG differed according to production stage. Skeletal-related body traits were more strongly associated with BW during the growing period, whereas nutrient intake variables became increasingly important during fattening. Integrating body measurements and nutritional information may support stage-specific growth monitoring and provide preliminary information for nutritional management in Hanwoo production systems.