Data as the stethoscope of the future
How often and for how long does a cow eat? Which chickens in a coop display problem behaviour such as feather pecking, and why? Sensors and cameras allow the behaviour of animals to be monitored ever more effectively. This produces an impressive amount of data and important insights into the influence of housing, animal management and genetic factors on animal welfare. However, there are also tough challenges. Do you really know what you are measuring? Who does the data belong to? And will the vet of the future have to be a data expert?
Sensors are not a new phenomenon in livestock farming. In dairy farming in particular, sensors are used to measure milk production, fertility, feeding behaviour and lying behaviour. ‘The behaviour of animals says a lot about their well-being. With new camera systems and smart data integration, we can monitor and explain this more effectively,’ says Bas Rodenburg, Professor of Animal Welfare at Utrecht University and Professor by Special Appointment at Wageningen University.
‘In practice, the use of sensors in the poultry and pig farming sectors is still in its infancy. Fitting sensors to the necks or legs of chickens and pigs is difficult: large numbers of relatively small animals are involved, so it’s not practically or commercially viable. With cameras, you can automatically monitor the behaviour of individual animals, as if you were observing them yourself 24 hours a day.’
‘There’s a wealth of untapped data
Data sharing
There is room for improvement in dairy farming too, says researcher Miel Hostens, who works with data and sensors for cattle at the Faculty of Veterinary Medicine. ‘Digitisation holds a great deal of promise, but it has yet to be fulfilled in practice. By no means all the sensors have been independently tested, so the reliability of the data is open to question.’ Hostens is working on solutions, such as the development of ISO certification for sensors.
A Billion observations
How many steps does a cow take per day, how often and how long does a cow eat and lie down? PhD candidate Peter Hut uses sensors in his research on cattle. To date, he has analysed over a billion observations to determine which factors have the most impact on well-being and health.
Prevention of disease
‘Lameness is a common problem in cows, which are forest animals by nature. Our analysis shows that cows that are lame and have lost weight are much more likely to eat less often and for less time. The question then is: does a cow become lame and lose weight as a result? Or will a cow stand up more due to a lack of comfort in the stable, for example, which results in painful hooves and a reduced appetite? With the data from sensors, we can move from treating problems to early detection or prevention of disease.’
Sensor manufacturers are often reluctant to release their algorithms. Hostens: ‘Clearly, there are major commercial interests at stake – but if we really want to help the sector move forward, we must find a way of sharing and analysing information. I’m confident that we’ll be able to do that.’ Hostens is supported by , director of , a company that supplies a platform on which several flows of sensor data at a farm are combined in a dashboard. ‘There’s only one owner of the data and that’s the farmer. If they see the benefits of sharing data, they will do so.’ Ten Cate is enthusiastic about the promising new camera systems and is therefore a partner in the research. ‘We’d love to include behavioural data in our dashboard, as a key predictor of well-being and health.‘
More research without laboratory animals
Under the umbrella of the research project IMAGEN (part of NWO’s Perspectief programme), Rodenburg and colleagues are looking for technical solutions to combine large data sets and algorithms, together with data experts from of Technology and breeding and behaviour experts from . They are doing this in a way that is workable for all parties. This is also important for research. Hostens: ‘There’s a wealth of untapped data. This will allow us to conduct far more research without the need for new laboratory animals.’
The data vet
With the advent of sensors, the farm vet will increasingly become a welfare adviser. Hostens and Rodenburg see the faculty’s new building plans as a good opportunity to pay more attention to data skills in the degree programme. Hostens: ‘The faculty’s farm De Tolakker will be a real innovation hub, where we will conduct research on data integration together with the sector, and where students will gain hands-on experience with using sensors in coops and stables.’
Insight into problem behaviour in the coop
‘It’s really a new way of looking at things. You have to see it as the difference between a car that uses a laser to estimate how much parking space you have left, or a completely self-driving car that can overtake other vehicles on the motorway.’ This is how PhD candidate Arjen van Putten describes the difference between existing sensors in the poultry farming sector and the camera system he is working on – first in a small test facility in Utrecht, soon at the Het Rondeel test coop of partner . ‘We look in detail at the behaviour of individual animals and investigate what causes problem behaviour such as feather pecking. Besides housing and management, we also take genetic factors into account. Together with , we explore how breeding can contribute to healthy behaviour.’