New insights into the association between lameness, behavior, and performance in Simmental cows

Area of Science:

• Animal Science
• Veterinary Medicine
• Dairy Production

Background:

• Lameness is a significant welfare and economic issue in dairy cows.
• Accurate prediction of lameness is crucial for timely intervention.
• Existing lameness detection models have limitations in utilizing behavioral and performance data.

Purpose of the Study:

• To evaluate the ability of automated performance and behavior parameters to predict lameness in Simmental cows.
• To identify key factors influencing the association between cow behavior, performance, and lameness status.

Main Methods:

• Collected 1-year of automated data on performance (milk yield, live weight, feed intake) and behavior (feeding patterns, activity, lying duration) from 100 Simmental cows.
• Assessed lameness weekly using locomotion scoring, classifying cows as lame (score 4-5) or non-lame (score 1-3).
• Utilized elastic net regression to analyze 25 daily parameters for predicting lameness status.

Main Results:

• The final model achieved high prediction accuracy (AUC = 0.91), with excellent sensitivity (0.94) and good specificity (0.81).
• Key predictors of lameness included fewer, shorter feeding meals and decreased feed intake per meal.
• Milk yield and lying duration were significant only through interactions with other factors, such as feed intake and feeding duration.

Conclusions:

• Automated monitoring of feeding behavior, specifically meal frequency, duration, and intake, offers a highly accurate method for predicting lameness in dairy cows.
• The interplay between intrinsic factors like feed intake and behavioral patterns is critical for understanding lameness prediction.
• These findings provide deeper insights into the complex relationships influencing lameness, improving the utility of behavioral and performance data for detection models.