Development of dry matter intake equations of dairy cows during dry period

In this study, four different models incorporating animal-related factors, blood biochemistry, and climate variables were developed to predict dry matter intake (DMI) in prepartum cows, and their performances were compared with reference models. The research was conducted using 17 dry-period Holstein cows. The developed models included independent variables such as body weight, shrunk body weight, metabolic body size, body condition score (BCS), calf birth weight, rectal temperature, blood biochemistry parameters (P, Mg, Ca, blood urea nitrogen, total protein, albumin, AST, GGT, glucose), and environmental factors (ambient temperature, relative humidity, wind speed, wet-bulb temperature, and temperature-humidity index). Model performances were analyzed using statistical criteria such as R², mean squared prediction error (MSPE), mean bias (MB), accuracy coefficient (ACC), concordance correlation coefficient (CCC), and Akaike Information Criterion (AIC). According to the results, MODEL 4, which included all variables, had the highest coefficient of determination (R²=0.637), indicating that DMI was significantly affected by body weight, serum Ca concentration, and wind speed. Compared to the reference models, the highest performance was observed in the NASEM (2021) and NRC (2001) models. To improve model performance, validation with larger datasets and testing in different geographical regions are recommended. The findings of this study may help optimize cow feeding programs, ultimately improving both animal welfare and productivity.