Elaine (Lenny) van Erp-van der Kooij has an MSc in Animal Science from Wageningen University and a PhD in Veterinary Science from Utrecht University in the Netherlands. Her PhD research was on coping behaviour in pigs. She is a professor in Animal Welfare and Technology at HAS green academy in the Netherlands and Visiting Professor of Practice in Precision Livestock Farming at Harper Adams University in the UK. Her main focus is improving the health and welfare of animals using sensor technology and data science. Projects and studies include AI and sensor applications in farm animals, horses and companion animals.
Lenny van Erp-van der Kooij
Professor (UAS) in Animal Welfare and Technology HAS green academy, university of applied sciences ‘s-Hertogenbosch, the Netherlands
AI and Sensors in Swine Farming: A Biological Approach
Artificial Intelligence (AI) and sensor technologies enable continuous, objective monitoring across all stages of swine production and are central to Precision Livestock Farming (PLF). Their main value lies in detecting biologically driven risks early enough for effective intervention. For sows, wearable sensors such as accelerometers quantify posture changes and activity patterns. They can predict farrowing up to 12 hours in advance by identifying nest-building behaviour and support early detection of lameness. Vision systems track lying behaviour and thermoregulation, while Infrared Thermography (IRT) provides non-contact assessment of skin temperature for early signs of fever. For neonatal piglets, camera-based analysis detects hypothermia, low mobility, and crushing risk. Acoustic monitoring identifies distress calls, enabling rapid support for low-vitality piglets. In weaned piglets, sound-based systems using neural networks detect respiratory distress, while cameras track growth, posture, and activity to flag early disease or pen fouling. IRT can identify temperature deviations associated with infection. For fattening pigs, vision technologies estimate body weight and monitor lying behaviour, aggression, and thermal discomfort. Acoustic algorithms classify coughs, supporting early detection of respiratory outbreaks. These systems provide earlier biological insight and more targeted intervention. Yet wider implementation still requires robust external validation and economically feasible deployment. Swine veterinarians and production specialists remain essential: they must interpret sensor alerts within normal physiological variation, ensure high sensitivity and specificity, and collaborate with engineers to define biologically meaningful behaviours for model development. Their expertise is key to translating automated data streams into improved welfare, reduced medication use, and more precise herd management.