Technology advancements—specifically machine learning and artificial intelligence (AI)—are transforming every industry, including equine veterinary medicine. From improving diagnostic accuracy to enhancing treatment protocols and streamlining administrative tasks, AI has the potential to be an invaluable tool for equine veterinarians.
In addition to using AI to drive efficiency and leverage data-driven decision-making, veterinarians are open to employing it to give horse owners what they want. Like it or not, horse owners are familiar with the technology. It’s part of their work, their entertainment, and their daily life habits. While the equine industry has often lagged behind in technology adoption, the trend is changing. Equine veterinarians, however, have an advantage over other professional industries: Clients can’t simply go to the big box store, buy the tech, and take a DIY approach.
“People are scared of AI taking over the human brain, but I think we need to be a little more open-minded,” says Ludovica Chiavaccini, DMV, DES, MS, a clinical associate professor of anesthesiology in the University of Florida’s College of Veterinary Medicine. “It’s like pretending to still do math with the abacus. AI will not replace veterinary care but will allow for early [detection] and a bigger picture of the horse.”
While AI will not replace veterinarians, if a client asks and their practitioner refuses to talk about it, the client might start looking for someone who will. At the same time, client conversations involving AI should explain that the technology is an enhancement of, not a replacement for veterinarians’ expertise.
“Clients should be aware of the fact that the AI only serves as an additional tool in combination with a complete examination of a veterinarian,” says Anna May, FTA Internal Medicine for Horses, DECEIM, a professor at the Equine Clinic at Ludwig Maximilian University of Munich, Germany. “It should be emphasized that it may only make the examination better because there is an additional ‘eye’ looking at their horse.”
Research related to applications of AI and machine learning for veterinary medicine is growing rapidly. Early results are providing exciting possibilities for enhanced care. However, it’s critical to be aware of limitations in current studies.
“As an editorial board member and reviewer for several AI-related journals, I am skeptical too of the reported results in the literature, but not of the AI potential,” says Mohammad Fraiwan, from the Jordan University of Science and Technology and co-author of Using Artificial Intelligence to Predict Survivability Likelihood and Need for Surgery in Horses Presented With Acute Abdomen (Colic).1 “The literature contains so many fake results, incorrect evaluations, and missing details. Furthermore, most of the results in the literature are reported under controlled circumstances such as photos taken under certain light intensity and without irrelevant elements.”
Nonetheless, Fraiwan believes AI has great potential. However, it needs large datasets to be reliable, which are mostly available to large corporations and governments (e.g., the military).
“The more AI applications find their way into our daily life, the less skeptical people will become,” he says. “In addition, controlled trials of AI software should convince skeptical people.”
3 AI Uses in Veterinary Medicine
The potential for AI in equine medicine is limitless. More experimentation, research, and refinement are necessary to understand the best applications and ensure accuracy. But it’s continually evolving, and this article includes a brief overview that barely scratches the surface of possible uses.
1. AI for Predicting Colic Surgery Survivability Rates
Colic is among the most common conditions practitioners see and causes significant concerns and expenses for horse owners. Often, horse owners must make quick decisions to invest in surgery in situations where survivability odds are unclear. Artificial intelligence and machine learning show promise for enhancing the predictability of survival rates to help guide this decision-making.
Fraiwan saw an opportunity to study a machine learning model to determine how accurately it could predict colic surgery survivability rates based on 606 records.
Twenty-five parameters were recorded for each case using the standard guidelines for cases of abdominal colic in horses. These include age, sex, breed, history of previous colic, response to analgesia, rectal findings, and more.
“Some variables were categorical with two or more categories, while others were continuous numerical values,” he explained. “Furthermore, the diagnosis was included in the data along with what was the outcome of the clinic visit, such as whether surgery was performed or not, and whether the horse was euthanized or not.”
Study results revealed that the average accuracy for predicting survivability likelihood was 85%, while the accuracy for predicting the need for surgery was 76%.
“Machine learning models, if developed properly using large datasets, can have great potential in aiding the diagnosis and prognosis process,” he said. “They can greatly help expedite the work of clinicians, eliminate errors, and reduce the need for specialized personnel.”
2. AI-Assisted Pain Assessment
Over the past 20 years, researchers have developed detailed ethograms or catalogs of facial expressions and the Horse Grimace Scale to identify pain in horses. The first pain-specific ethogram emerged in 2005, with further validation in 2014, expanding to acute (e.g., castration) and chronic (e.g., laminitis) pain assessments, says Chiavaccini.
Inspired by a graduate student’s love of goats, Chiavaccini led a research team that filmed goats’ facial expressions during pain and comfort and then fed the data into an artificial-intelligence-based model that learned to distinguish a goat’s pain by its face alone. But Chiavaccini is more interested in horses than goats and started looking to apply AI to the recognition of pain in horses.
In 2024, she and a team of researchers published “From Facial Expressions to Algorithms: A Narrative Review of Animal Pain Recognition Technologies”2 to address the drawbacks of human subjectivity when identifying facial expressions and pain recognition. The team leveraged AI algorithms to “construct sophisticated models capable of analyzing diverse data inputs, encompassing not only facial expressions but also body language, vocalizations, and physiological signals, to provide precise and objective evaluations of an animal’s pain levels.”
“When we presented our project to a veterinary conference of anesthesiologists, [some said,] ‘I don’t trust this replacing my clinical experience. When I assess pain in an animal, I also take into consideration heart rate, respiratory rate, and this and that,’ which is true,” she says. “Our model was far from perfect—it was just the beginning. The idea is to also train the model with all the data, so potentially, it can give you a better prediction, and that could be an aid, especially for owners, trainers, or help for the veterinarians, not necessarily replace them.”
Future progress depends on larger datasets and closer collaboration between engineers and veterinarians to ensure AI models are both technically robust and clinically relevant.
3. AI-Aided Differentiation of Ophthalmic Diseases
Uveitis is an emergency that can severely damage a horse’s eye and result in blindness. Therefore, accurate and timely diagnosis of the disease is important, says May. Her research team hypothesized they could save more horses’ eyes if they had a reliable tool to diagnose the disease.
“We had many photos of equine eyes and wondered if an AI tool could be developed based on pictures,” she says. “The eye itself is a small and confined part of the horse’s body, and many findings can be seen just by looking at the eye from the outside.”
The research team, which is familiar with eye conditions some veterinarians might have never seen, trained the AI tool with many features of equine ophthalmic diseases. Their goal is to connect pictures with clinical data.
“We lack evidence-based medicine in equine medicine and, in the future, the AI can be helpful to properly define disease categories,” she says. “Equine ophthalmology is a peripheral area of equine medicine, and AI can help inexperienced vets with diagnoses and guide them to case-relevant information.”
The study results showed that the AI Equine A-Eye web app is reliable in diagnosing uveitis and other diseases in photos. At this point, it is not commercially available, but the results have shown its potential in supporting veterinarians to make a diagnosis of equine ophthalmic disease.
“In the version used in the study (May et al. 2022 and Scharre et al. 2024), the accuracy rate for the validation data was 96.7%,” she says. “[This tool] can support in areas without veterinary emergency coverage, support inexperienced examiners who are not familiar with an equine eye examination, and create better communication between owners and veterinarians.”
Veterinary AI Applications On the Horizon
There are countless applications for AI in equine practice, some of which have been proposed in the literature, says Fraiwan. They include the analysis of equine images, such as the horse’s emotional state, and pain, histopathology applications, blood analysis applications, and audio applications to analyze horses’ behavior or improve their well-being.
As an anesthesiologist, Chiavaccini is eager to see how AI and machine learning can help predict which horses will recover poorly from anesthesia. Between 30% and 40% of anesthesia accidents happen during recovery, with large horses at greater risk.
“We know that it seems like larger horses are more at risk and that hypoxemia may be a factor,” she says. “While we suspect, ‘This horse has a 40% increased risk to have a bad recovery,’ we don’t have a good prediction model to know that yet. What if we put all this information in an artificial neuronal network and potentially simulate each case based on the information we collect at the end of the procedure, including whether or not the horse recovered well? Could the computer or the artificial neuronal network more correctly predict it?”
Other AI research and experimentation includes lameness detection and gait analysis. Lameness can be frustrating and expensive to detect and treat. Researchers are exploring ways to use AI technology to detect even subtle gait changes in numerous disciplines and situations. They’re using a variety of technology tools to collect and analyze a horse’s movement in real time to help spot asymmetries and abnormalities earlier.
Another area of interest among researchers is the use of AI to detect equine asthma. In one study, “Evaluating Asthma in Equines with Video Recordings,” researchers presenting at the 2024 EPIA (Portuguese Conference on Artificial Intelligence) conference shared they had an 89% accuracy rate using AI to identify asthma in horses through video surveillance (Gomes et al.).
May adds that, in her perspective, the most exciting benefit of using AI in is it will help with repetitive tasks, so veterinarians have more time to devote to patients and their owners.
“In order for professionals and customers to gain trust, the AI decisions need to be elaborated and explained,” Fraiwan says. “Much like the vet explains his rationale to the client, they should be able to elaborate/convey the AI decision to the client. For example, an AI algorithm may claim that a histopathology sample is positive (e.g., cancerous). Traditional AI did not explain the reasons for the decision (i.e., a black box). On the other hand, explainable AI aims to produce an explanation (e.g., marking of the tumor cells in the image), which would allow the professional to verify the decision if need be.”
Take-Home Message
AI and machine learning have the potential to revolutionize equine veterinary care through enhanced diagnostics, allowing for earlier disease and lameness detection and more precise treatments. More research is needed to confirm and validate the accuracy and efficacy of these technologies in veterinary medicine. However, the potential is widespread and becoming visible not only through studies in equine medicine but also work done in other species—and even in unrelated industries—that can be transferred to horses. Approaching AI and machine learning with an open mind and human critical thinking to identify flaws and inaccuracies presents opportunities to improve animal health and welfare. The possibilities for increased efficiencies can grant veterinarians much-needed extra time in the day.
References
- 1. Mohammad, F.A. and Abutarbush, S.M. Using Artificial Intelligence to Predict Survivability Likelihood and Need for Surgery in Horses Presented With Acute Abdomen (Colic). Journal of Equine Veterinary Science. Vol. 90 (July 2020) https://www.sciencedirect.com/science/article/abs/pii/S0737080620300642.
- 2. Chiavaccini, L., Gupta, A. and Chiavaccini, G. From facial expressions to algorithms: a narrative review of animal pain recognition technologies. Frontiers in Veterinary Science. July 2024. https://pubmed.ncbi.nlm.nih.gov/39086767/
- 3. May, A., Gessell-May, S., Müller, T. and Ertel, W. Artificial intelligence as a tool to aid in the differentiation of equine ophthalmic diseases with an emphasis on equine uveitis. Equine Veterinary Journal. Oct. 2021.
Related Reading
- Equine Pain Assessment Tools and Techniques
- Researchers Use Artificial Intelligence to Identify New Antibiotic
- The Business of Practice: AI Scribes for Veterinary Records
Stay in the know! Sign up for EquiManagement’s FREE weekly newsletters to get the latest equine research, disease alerts, and vet practice updates delivered straight to your inbox.
