Classification of lameness in dairy cows using automatic milking system data and body condition score with machine...

Area of Science:

• Animal Science
• Veterinary Medicine
• Dairy Production Systems

Background:

• Lameness significantly impacts dairy cow welfare and herd productivity.
• Early detection of lameness is crucial for timely intervention and management.
• Automatic milking systems (AMS) generate extensive data that may aid in lameness assessment.

Purpose of the Study:

• To investigate the influence of animal traits (parity, body condition score - BCS) and environmental factors on locomotion score (LS) in lactating dairy cows.
• To assess the impact of lameness on milking parameters using data from AMS.
• To develop and evaluate a machine learning model for lameness classification in dairy cows.

Main Methods:

• Data collected biweekly on locomotion score (LS) and body condition score (BCS) from 323 lactating cows over 7 months on 3 Italian farms with AMS.
• Analysis of 42,569 AMS observations using linear mixed models to correlate LS with milking parameters.
• Application of multiple correspondence analysis and an extreme gradient boosting machine learning model for lameness classification.

Main Results:

• Cows with parity 3 or greater and lower BCS exhibited significantly higher LS.
• Severely lame cows showed reduced daily milk yield, fewer milkings, longer milking duration, and delayed milk flow, especially in rear quarters.
• The machine learning model achieved 92% balanced accuracy in classifying cows into nonlame, mildly lame, or severely lame categories.

Conclusions:

• Body condition score (BCS) and parity are significant predictors of lameness in dairy cows.
• AMS data, particularly milk flow and milking frequency, are valuable indicators for lameness detection.
• Integrating AMS data with BCS and machine learning provides an efficient, equipment-free approach for lameness monitoring and early detection in dairy herds.