Spotting Sick Cows by Analyzing Their Health Data Patterns

Jim Crocker
21st February, 2024

Spotting Sick Cows by Analyzing Their Health Data Patterns

Domestic Cattle (Bos taurus)

Photo adapted from: Sagnik Dutta Roy / CC BY (Source)
Maintaining udder health is crucial for efficient dairy farming. A key indicator of udder health is the somatic cell count (SCC) in milk – elevated levels suggest inflammation, often due to mastitis, an infection of the udder. Detecting mastitis early is important to minimize its impact on milk production and animal welfare. The University of Copenhagen researchers recently conducted a study[1] to develop a method for identifying abnormal SCC patterns in dairy cows using routinely collected data, potentially leading to earlier mastitis detection. The study focused on Holstein dairy cows from eight conventional herds in Denmark, analyzing data collected between 2010 and 2020. Data was sourced from the Danish milk recording system and the Danish Cattle Database, which routinely records SCC for individual animals across multiple milk recordings throughout the year. The researchers selected data from 13,996 cows, prioritizing herds with substantial data availability to ensure robust analysis. Traditionally, SCC levels fluctuate throughout a cow’s lactation cycle. They typically rise shortly after calving, then decrease, and slowly increase again towards the end of lactation. The researchers aimed to define what constitutes a “normal” SCC pattern for each individual cow, and then identify those deviating from this norm. To achieve this, they used mathematical models – specifically, functions developed by Wilmink and Wood originally designed to model milk yield – to describe the typical SCC curve over the course of lactation. These models were fitted to each cow’s SCC data, transforming the SCC values using a logarithmic scale for improved statistical analysis. The core of the method lies in identifying outliers based on the ‘mean squared residuals’ (MSR). Essentially, the MSR measures the difference between the actual SCC data and the SCC predicted by the model. A high MSR indicates a poor fit, suggesting the cow’s SCC pattern is unusual. By identifying cows with consistently high MSR values, the researchers could pinpoint those with SCC curves that deviate significantly from the expected “normal” pattern. The Wood’s style function consistently provided a better fit to the data than the Wilmink function. This approach builds upon previous research demonstrating the value of monitoring SCC for mastitis detection. For example, studies have shown that different pathogens causing mastitis result in distinct SCC patterns[2]. Staphylococcus aureus infections tend to cause persistently high SCC, while E. coli infections often lead to a rapid spike and subsequent decline in SCC. The current study doesn’t directly identify specific pathogens, but rather flags deviations from an individual cow’s expected pattern, which could then prompt further investigation to determine the underlying cause – including pathogen identification. Furthermore, the discovery of regularly fluctuating SCC patterns in cows milked by automatic milking systems[3] highlights the complexity of SCC data. While the current study focuses on identifying deviations from a cow’s own baseline, understanding these inherent fluctuations is important for refining the accuracy of abnormality detection. The “Dairy Brain” concept, which aims to integrate various data streams for improved dairy farm management[4], could readily incorporate this new method. The ability to automatically identify cows with abnormal SCC patterns from routine registry data aligns with the Dairy Brain’s goal of providing real-time, data-driven decision support. This could allow farmers to proactively address potential mastitis cases, potentially reducing treatment costs and improving herd health. The study’s findings suggest a pathway towards a more automated and efficient system for monitoring udder health in dairy herds.

AgricultureBiotechAnimal Science

References

Main Study

1) Using registry data to identify individual dairy cows with abnormal patterns in routinely recorded somatic cell counts.

Published 21st February, 2024

https://doi.org/10.1016/j.jtbi.2023.111718

Related Studies

2) The effect of pathogen-specific clinical mastitis on the lactation curve for somatic cell count.

Journal: Journal of dairy science, Issue: Vol 85, Issue 5, May 2002

3) Regularly fluctuating somatic cell count pattern in dairy herds.

https://doi.org/10.3168/jds.2020-20063

4) Symposium review: Real-time continuous decision making using big data on dairy farms.

https://doi.org/10.3168/jds.2019-17145


Related Articles

An unhandled error has occurred. Reload 🗙