Evaluation of different feed intake models for dairy cows

Area of Science:

• Animal Science
• Dairy Nutrition
• Statistical Modeling

Background:

• Accurate feed intake prediction is essential for optimizing dairy cow diets and performance.
• Existing models vary in complexity and data requirements, necessitating evaluation for practical application.

Purpose of the Study:

• To assess the performance of different feed intake prediction models using diverse data from lactating cows.
• To evaluate model robustness when incorporating milk yield data, considering practical data availability.

Main Methods:

• Utilized a dataset of 2,161 individual observations from 24 change-over trials across Nordic countries.
• Employed residual analysis with mixed and simple model regression to evaluate prediction accuracy.
• Compared models using observed, previous period, and average milk yield data.

Main Results:

• Models showed varying biases and prediction errors, with the Finnish and Scandinavian models demonstrating strong performance.
• Prediction errors were generally higher when using previous period data compared to observed milk yield.
• The inclusion of current dietary factors significantly impacted milk yield, affecting intake predictions.

Conclusions:

• Milk yield is a valuable input for feed intake prediction models when aiming to maintain current production levels.
• Models incorporating current milk yield can lead to substantial errors if the optimal diet differs from the current one.
• Model selection should consider data availability and the specific goals of ration formulation for dairy cows.