Artificial intelligence’s role in veterinary medicine

Image: Kaikoro/Adobe Stock

Veterinary clinics and professionals across the globe are beginning to experiment and test how artificial intelligence (AI) can help them take better care of their patients while striving to improve efficiency. As the benefits of AI emerge along with multiple products, some risks are beginning to come to light.¹

During his specialty symposium lecture at the American College of Veterinary Medicine (ACVIM) Forum in Philadelphia, Pennsylvania, Eli B. Cohen, DVM, DACVR, clinical professor at NC State College of Veterinary Medicine in Cary, North Carolina, broke down the ethical and legal dangers that veterinary professionals should consider if they are looking to bring AI into their clinic.

“We need to make sure that the folks who are developing these algorithms products and the end users do have a conscience because the AI does not. And if it does, if you've seen any movies, that doesn't always end well for us. Because what it means might not always align with what we need. So, we can't afford to be indifferent,” said Cohen.

When it comes to treating patients, the use of AI can be a benefit but also a danger because it can create an unintentional bias. During his lecture, Cohen presented an example of a dog that was brought into the veterinary clinic for treatment. The canine was unable to walk and had a breakthrough fracture on its leg. After taking radiographs, the clinic used AI technology to conclude the canine’s treatment needs. However, Cohen gave a warning of what could happen when only relying on AI technology and not using your hands and skills alongside it.¹

“Maybe we didn't circle back and finish our neurology exam or once the patient was out of shock and we don't realize that this patient also has a luxated back. Maybe that's why they're not walking. Then to take a step further, maybe the reason they got whacked by a car in the first place is because they have a brain tumor, and they want it out on the street,” explained Cohen. He noted that these are all ways a patient could have a satisfactory treatment that’s wrong in the medical record. “The more something is said or written or in a record, it picks up steam and becomes this monolith of a diagnosis and can impact subsequent decisions and other folks that pick up that case,” he continued.

As AI helps to diagnose an issue a pet is facing at the time, it may prevent veterinary professionals from discovering conditions that may not be as noticeable. AI is a tool that should be used in tangent with professionals completing full exams, not in replace of them, according to Cohen. Veterinary professionals who are interested in trying AI at their practices should do their research and look into the upcoming AI technology tools, because there is not a regulation out there for them and this can have harmful reactions, he added.¹

“There are no requirements from the FDA for premarket approval for medical devices intended for animal use. Okay, so that's funky, right? It basically means you can bring any AI product to market and use it on animals…There is no regulation, there's no checking, there's no validation, you can bring anything to market, you don't have to say anything about how you trained it or how accurate it is, you can just bring it to market,” said Cohen.

“Now, these animal medical devices and aids are subject to the provisions that relate to misbranding and adulteration. Particularly if labeling or marketing is false or misleading, so that is a place where things can be brought to task if someone chooses to. But otherwise, you can be just off to the races,” concluded Cohen.

Reference

Cohen E. Ethical and Legal Implications of AI in Veterinary Medicine. Presented at: ACVIM Forum Specialty Symposium; Philadelphia, Pennsylvania. June 14, 2023.