A decision-making tool developed by DTU Vet and DTU Compute aims to boost production and animal welfare.
By Anne Lykke
When a dairy cow becomes sick, there comes a point when it makes more financial sense to take it out of production than run the risk of it infecting the rest of the herd. But how do you decide the cut-off point for when the cow is productive or poses a health risk?
The new iCull project seeks to address this problem. The project aim is to create a model to determine when best to withdraw a sick dairy cow from production, balancing productivity and state of health—and incorporate the findings into an app.
The benefits include improved economy for milk producers, improved animal welfare and reduced carbon footprint per litre of produced milk—thus helping the environment.
Every year, 150,000-200,000 cows from the approximately 3,700 Danish dairy herds are withdrawn from production owing to poor milk yields or illness.
The decision-making tool is specifically developed to assess the risk of paratuberculosis—a chronic bacterial infection that causes diarrhoea in cattle and other ruminants. However, it should be possible to refine the tool to identify other bovine diseases such as mastitis.
The project employs advanced simulation models to predict Daisy the Cow’s future yield and risk of infection.
“DTU Vet possesses extensive expertise in epidemiology simulation, paratuberculosis and production conditions—but when things become really programming-intensive, it’s great to have DTU Compute on your side,” says Senior Scientist Tariq H. Halasa from the Section for Immunology and Vaccinology at DTU Vet, which is partnering the project.
At DTU Compute, Associate Professor Lasse Engbo Christiansen also sees numerous advantages to inter-departmental cooperation:
“We are delighted to collaborate with DTU Vet on these types of projects that occupy a grey zone between biology, large data sets and modelling. Often, the challenge is getting your dream model to work fast enough. At DTU Compute, we can optimize technical programming to ensure the model works efficiently. This is incredibly valuable because it allows you to try out various scenarios faster,” he explains.
Measuring input and output
Two postdocs—one in each department—are working closely on the project to produce a smartphone or computer app that can be used by milk producers, vets and other advisers. The tool will use input, including the cow’s milk, feed intake, reproduction and udder health as a basis for calculating the specific animal’s so-called future value—but other parameters such as blood and milk samples, feed type and the age of the cowshed will also be included to see whether they reveal anything about health and illness.
The idea is to constantly exploit the collected data.
Better economy for farmers
Currently, paratuberculosis testing is carried out using antibodies in the milk approximately once every three months. The vet or adviser then decides which animals to take out of production. However, as herds grow larger, it is becoming increasingly difficult to assess individual animals in this way.
“We would like to reach the point where the vet or adviser can use the app to make a decision based on lots of calculations he or she cannot make alone. As things stand, farmers risk waiting too long before withdrawing an animal from production. It can then go on to infect the rest of the herd, causing disproportionate losses over the long term,” says Tariq H. Halasa.