Can a horse talk and tell you when it’s in pain?
Of course not, unless it was Mister Ed, the palomino gelding star of the popular 1960s TV sitcom that could talk, but only to its owner, Wilbur Post – an architect who moved into a suburban home with a barn in the backyard and was the only person worth talking to.
Ed would make phone calls, offer witty, sarcastic commentary on human affairs, or cause mischief, leaving Wilbur in bizarre situations and trying to explain his horse’s behavior to everyone without revealing Ed’s secret.
Dr. Marcelo Feighelstein, head of the Artificial Intelligence Systems Engineering Program at Tel Hai University of Kiryat Shmona in the Galilee (which has just taken a step up from being Tel Hai College), is not Wilbur Post – but he did want to discover when horses and other animals suffer from pain.
His new AI framework, known as SHIC-XE, has been developed to detect signs of pain in horses via video analysis while providing stable, anatomically consistent explanations for its decisions.
Feighelstein led the collaboration that included Prof. Anna Zamansky and the Tech4Animals Lab, Prof. Ilan Shimshoni of the University of Haifa, students Omer Bibi and Ofer Rosenbaum from Haifa’s Technion-Israel Institute of Technology, and researchers from the University of Bern, the University of Milan, the University of São Paulo, and Newcastle University.
It was published in the prestigious International Journal of Computer Vision under the title “SHIC-XE: Viewpoint-Invariant Explainability via Dense 2D-3D Correspondences: an Application to Equine Pain Recognition.”
The research found a solution to an enduring challenge in both veterinary and human medicine: how to identify pain and distress in individuals who cannot communicate what they are experiencing.
People who could benefit include newborns, people in intensive care who are sedated and ventilated, and victims of dementia, stroke, progressive aphasia, head injuries, tumors, or serious infections who are unable to speak. It could also analyze neurological movement disorders and assist with clinical and surgical video analysis.
“For a long time, doctors thought that newborns didn’t feel pain, so they didn’t take measures to prevent their suffering,” he noted.
For the first time, according to the team, these AI-generated explanations can be quantitatively compared with expert assessments, representing a significant breakthrough in the field of explainable artificial intelligence and an important step toward AI systems that can be trusted in real-world clinical and healthcare environments.
Traditional methods use 2D visualization techniques such as saliency maps – visual heat maps that highlight the most important parts of an image for a human viewer or a machine learning model. But when applied to video data, subject and camera motion produce unstable, flickering maps that cannot be temporally aggregated in meaningful ways.
“This limitation is critical in medical and veterinary settings, which demand biologically grounded explanations intuitive for experts. Horses are particularly challenging in the context of pain as they are known to hide pain signals in human presence,” the chief researcher said.
AI supporting the doctors
Animals can't express pain in words, and even experienced veterinarians and caregivers can struggle to recognize signs of suffering, said Feighelstein, who earned his doctorate in computer science and information systems at the University of Haifa and focuses on AI for animal welfare, animal emotion recognition, agriculture and dairy farming, and AI applications.
Many animals, from dogs to horses, try to hide their suffering so they won’t be vulnerable to attack. Feighelstein’s research bridges that communication gap by using AI to interpret facial expressions, body language, and movement patterns.
Over the years, he and his team have developed AI-based tools capable of recognizing pain and emotions in cats, dogs, rabbits, sheep, cattle, and now horses.
“Working on dogs is more difficult because there are hundreds of species – from Shih-Tzus and Pomeranians with short noses to collies, greyhounds, and dachshunds – that have long noses and different facial features.”
His motivation is “to build a bridge between humans and animals,” Feighelstein told The Jerusalem Post in an interview. “We want to give a technological voice to those who have no words, allowing caregivers and professionals to better understand their condition, their emotions, and their suffering.
Until now, methods to detect pain in animals often highlighted different areas of an image from one video frame to the next, creating unstable and difficult-to-interpret explanations. To overcome this limitation, the researchers developed SHIC-XE, a novel framework that projects the horse’s attention onto a fixed 3D representation of its face.
This, he said, creates a consistent anatomical explanation, even when the animal moves, changes its head position, or is filmed from different angles.
“We proved that the model not only reaches the correct conclusion, but also focuses on the anatomically relevant regions, especially the ears and cheek muscles, when making that decision,” he said. “This brings us closer to opening the ‘AI black box’ by showing not only what the model decides, but whether the evidence it uses makes sense to experts.”
When they began studying automatic pain and emotion recognition in animals, “the goal was to give them a voice,” Feighelstein continued. “The scientific question is not merely ‘Can AI detect pain?’ but ‘Can we trust an AI system when it tells us an animal – or eventually a human patient – is in pain, and can it show us why?’”
Asked what the SHIC-XE system does that existing AI systems cannot, he said that the new method “pays attention to more than an image of the face; it also works when the animal is moving in video. Existing systems could show attention in one frame and indifference in the next.”
There is no danger that vets would become too dependent on an AI pain detector, Feighelstein asserted.
“It will support the doctors and not replace them. It could eventually provide continuous monitoring, alerting a veterinarian when an animal’s condition changes even when nobody is watching it. The technology could be used on farms, in veterinary hospitals, laboratories, or animal transportation, but first we need to build an app from the model.
“We are now working on this. Now, it detects only the presence of pain, but eventually it could estimate its severity. We will work on this in the future. We also hope to determine where the pain comes from – inflammation, orthopedic, surgical, or something else.”