AI Breakthrough Lets Horses 'Talk' to Vets: The Revolutionary System Detects Hidden Health Issues

Researchers are using artificial intelligence (AI) to bridge the communication gap between horses and humans. By combining 3D motion capture technology with machine learning, a new modeling system offers veterinarians a powerful visual tool for interpreting equine body language—a key to detecting both physical and behavioral problems.

This innovative system, developed by researchers at KTH Royal Institute of Technology and the Swedish University of Agricultural Sciences (SLU), can reconstruct the 3D motion of horses from video footage using an AI-based model that analyzes the horse's posture and shape in detail.

The model is precise enough to help veterinarians identify subtle changes that might otherwise go unnoticed or be misinterpreted during an exam, such as alterations in posture or body weight. This groundbreaking study was published on the arXiv preprint server.

Named Dessie, the system uses disentangled learning, which separates different factors in an image to prevent the AI from being confused by background details or lighting variations. "Dessie is the first instance of disentangled learning applied to non-human 3D motion models," said Hedvig Kjellström, a Professor in computer vision and machine learning at KTH.

Elin Hernlund, an Associate Professor in biomechanics at SLU and equine orthopedics clinician, explained that Dessie would provide more accurate observations and interpretations of horses' movements, leading to earlier and more precise medical interventions. "It’s like getting critical information 'straight from the horse's mouth,'" Hernlund said.

Horses communicate their discomfort or pain through body language, and their gait can reveal whether they are compensating for pain. Hernlund notes, "With Dessie, we've created a digital voice for these animals, allowing them to break through the communication barrier and tell us how they feel."

In addition, the research team is training Dessie with images from a broader range of horse breeds and sizes. This will help them understand the genetic links to physical characteristics, providing deeper insights into the biological structure of these animals. "We're asking breeders to send us images to capture as much variation as possible," Hernlund added.