Artificial intelligence poised to transform veterinary care

It’s difficult to grasp the prevalence of artificial intelligence (AI) given its rapid and seamless integration into our lives. From a smartwatch tracking our sleep patterns and fitness levels to self-driving cars, AI-powered technologies promise to radically reshape our world.

Decades ago, visionaries questioned whether computers would one day have a place in human health care, helping physicians make more accurate diagnoses. Fast forward to January 2022: Johns Hopkins University announced that the Smart Tissue Autonomous Robot (STAR) had successfully performed laparoscopic surgery on the soft tissue of a pig multiple times, each time outperforming a human doing the same procedure.

To get a sense of how AI is transforming the animal health space, Cornell University College of Veterinary Medicine hosted the first Symposium on Artificial Intelligence in Veterinary Medicine (SAVY) April 19-21 at its Ithaca, New York, campus. Several AI-based projects were shared by speakers and during poster presentations along with discussions about the current challenges and future potential of this emerging technology.

Adopting AI in and outside the clinic

Artificial intelligence is a computer simulation of human intelligence processes, such as learning, reasoning, and self-correction, to solve a problem or perform a task. Instead of a brain, computers have algorithms, a series of step-by-step instructions for “thinking” about data inputs to achieve the desired goal. Machine learning is a subsection of AI where the algorithm isn’t given a set of instructions, but rather trained on data to make decisions or predictions on its own.

Like any technology, AI is intended to improve our lives, by solving complex problems, automating tasks, and improving efficiencies and decision-making.

Veterinary medicine was slow to embrace AI, but that is changing. On the first day of the symposium, Sebastian Gabor, cofounder and CEO of Digitail, a cloud-based practice management software (PIMS) that uses AI to increase productivity, presented findings from the first industry-wide survey on AI in veterinary medicine. Digitail worked with the American Animal Hospital Association (AAHA) to survey nearly 4,000 of its members about their attitudes on AI, if they use AI in their practices, and if so, how, Gabor explained.

Of the approximate 83% of respondents who reported familiarity with AI, nearly 30% of them said they already incorporate AI into their practices, either on a daily or weekly basis. The finding surprised Gabor. He explained that it indicates veterinarians scored high on the adoption curve, meaning they are quick to seize on new technology, including seasoned practitioners.

It turns out veterinarians of all generations, including those approaching retirement, are excited to learn about AI voice-to-text tools that quickly transcribe client conversations and incorporate the information into the patient’s medical record. “This technology allows veterinarians to stop doing something they don’t enjoy, like updating medical records by hand, and spend more time with patients, which probably explains why the adoption curve for veterinarians is higher than normal,” Gabor said.

The survey revealed that the reliability and accuracy of AI systems are the most prevalent concerns, with 70.3% of respondents highlighting them. Data security and privacy worries were listed by 53.9% of participants, followed by 42.9% citing lack of training and knowledge.

Outside the clinic, veterinary researchers are working with data scientists, statisticians, machine learning engineers, and technology experts to develop AI tools for use throughout veterinary medicine. Projects already underway include detecting early signs of lameness in sheep, forecasting the spread of Lyme disease, projecting the severity of an outbreak of porcine reproductive and respiratory syndrome virus, and developing rapid diagnosis and staging of canine myxomatous mitral valve disease.

Pet owners are seeing a growing number of AI-powered gadgets meant to ease some of the challenges of owning an animal. For instance, two companies, Petnow and iSciLab, are developing nose print recognition technology for dogs that could eventually replace microchip identification. Smart collars track a pet’s vital signs and activity in real time, alerting the owner to possible changes in the animal’s health such as seizures.

Scaling up the power of AI

Geert De Meyer, PhD, heads the data analytics area for science at Mars Petcare, whose numerous animal health holdings include Banfield Pet Hospital and BluePearl. During his keynote address at the Symposium on Artificial Intelligence in Veterinary Medicine (SAVY), De Meyer explained that he and his team of over two dozen researchers have developed AI-powered tools used throughout Banfield and BluePearl hospitals.

For instance, RenalTech, is a proprietary technology using AI in the early detection of feline chronic kidney disease (CKD). As De Meyer explained, RenalTech was created with a dataset drawing on the medical records of hundreds of thousands of cats, analyzing blood and urine data of cats before and after a CKD diagnosis. The result is a tool that can help to predict whether a cat will develop CKD within two years.

AI’s predictive capabilities are making it possible to personalize patient treatment, improving the chances of a successful outcome. During the symposium, veterinary oncologist Dr. Joseph Impellizeri spoke on how machine learning algorithms and live cancer cell analysis are being used to predict the efficacy of anticancer drugs for lymphoma in individual dogs.

Lymphoma is among the most common types of cancer in dogs. The traditional treatment protocol for lymphoma—cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP)—uses a mix of chemotherapies. The traditional treatment method doesn’t account for the individuality of each patient, Dr. Impellizeri said, and patient response varies. ImpriMed, a California-based startup, offers personalized treatment protocol that uses AI to predict the efficacy of more than a dozen drugs commonly used to treat canine lymphoma.

“There is clearly a need for more personalized medicine and the ability to assess whether or not a lymphoma has a better chance to be responsive to a certain drug would be very helpful,” Dr. Impellizeri said.

AI on the farm

Veterinary researchers are also discovering ways of using AI to improve efficiencies on the farm.

Dr. Jasmeet Kaler, an associate professor at the University of Nottingham School of Veterinary Medicine and Science in the U.K., delivered a keynote address during the symposium about how precision livestock informatics can improve animal production, health, and welfare.

Take a dairy operation as an example. Dr. Kaler explained that monitoring devices fitted to each cow and throughout the production facility provide real-time biological and behavioral data, such as feed intake, thermal imaging, and posture. Algorithms are constantly analyzing these inputs, flagging subtle deviations from baseline health and potentially the early stages of illness that might otherwise go unnoticed.

Dr. Kaler is overseeing the development of another rapid-detection tool for the farm, one that uses AI to identify early signs of lameness in sheep. “Lameness is one of the biggest health and welfare challenges around the world and in the U.K.,” she said.

As a prey species, sheep hide signs of lameness if they feel threatened by human observers, making early detection extremely difficult. Dr. Kaler and her research team are working with industry partners to rectify this problem by developing AI tools capable of identifying the early onset of lameness, including detecting behavioral indicators when the sheep is standing, lying, and walking.

The challenges for veterinary researchers wanting to collect food animal data to develop AI tools are greater than those in companion animal medicine, according to Dr. Kaler. Livestock producers are especially reluctant to open their farms to outsiders who want to record animals and employees in a production environment. Nor are producers keen on disclosing details about their business.

Predicting disease outbreaks

Population medicine is another field of veterinary medicine where AI’s analytical capabilities show promise.

Dr. Beatriz Martínez López, director of the Center for Animal Disease Modelling and Surveillance at the University of California-Davis School of Veterinary Medicine, uses algorithms, machine learning, and big data analytics to understand how foot-and-mouth disease, African swine fever, and other infectious diseases spread among animal populations.

“AI can significantly enhance prevention, early detection, and faster control of livestock diseases,” Dr. López said during her keynote presentation at the symposium.

She highlighted a report in Nature published in October 2023 describing a machine learning model designed by her laboratory to predict emerging infections in swine production systems throughout the production process on a daily basis. The model accounted for such variables as nearby farm density, piglet inventory, and wind speed and direction.

The model demonstrated a good ability to predict infections, Dr. López said, adding that veterinarians and producers can use these daily infection probabilities as a benchmark for preventive and control strategies on farms.

AVMA and technology

The AVMA has recently formed a Task Force on Emerging Technologies and Innovation to provide practical support and resources in this area of practice.

The task force has been charged with developing a strategy by which the AVMA can best support practitioners faced with the opportunities and challenges of emerging technologies. It will also suggest a charge for the task force and potential members for a future AVMA committee that will develop related policy and create resources to support veterinary practitioners in the effective and safe implementation of these technologies in veterinary practice.

A version of this story appears in the July 2024 print issue of JAVMA