Artificial intelligence has enormous potential to improve efficiency in vet practices, but there’s a way to go before the technology goes mainstream. By Angela Tufvesson
A pet owner has a sick pet. Instead of contacting the local vet practice, the first thing they do is enter their pet’s symptoms into an app, which assesses the seriousness of the symptoms. It then suggests one of three care options: have a video consultation with a vet through the app, make an appointment with their usual vet or, for symptoms that don’t turn out to be serious, do nothing at all.
This is Joii, a British app that uses artificial intelligence—or ‘AI’—to triage sick pets. Given the increased demand for socially distant pet care, it’s slated to launch in Australia later this year.
“Joii gives owners access to a free symptom checker where they go through a few instinctive steps where they tap on body maps and do things like upload photographs and video clips,” says veterinarian Dr Robert Dawson, co-founder of Vet-AI, which created Joii. “Then it tells them what they need to do as their next step.”
What is AI?
Apps like Joii use AI technology to ‘think’ and act like humans. These sorts of platforms can perform tasks or imitate behaviour that would normally require human intelligence like speech recognition, visual perception and problem solving.
Two key ingredients have transformed AI from the stuff of science fiction to an everyday reality. The first is the capacity of humans to program or ‘train’ the machines—so there’s no need to worry that computer intelligence will soon equal human intelligence. With what’s called ‘machine learning’, we give computers the ability to improve and ‘learn’ over time.
The second is data. Lots and lots of data. “Having a database or some system of collecting data is an essential requirement to build and use these systems,” says Dr Mehar Khatkar, a senior research fellow at the Sydney School of Veterinary Science at The University of Sydney. “Whenever there’s a possibility of a good data set, then it’s a good candidate for applying AI techniques.”
Dr Dawson says the information provided by pet owners is fed into Joii, which over time “builds up patterns and an understanding of the most likely causes of problems”. This means the app can do things like assess the severity of an injured dog’s paw and identify cases of arthritis. The more data it collects, the better it gets at assessing symptoms.
There’s enormous potential to harness AI to deliver higher quality care with greater efficiency in veterinary medicine. Dr Dawson says there are lots of opportunities to use similar AI technology to that used in human medicine in diagnostics—in particular, in scanning X-rays, CT and MRI scans for abnormalities.
Another opportunity lies in the ability of AI to streamline customer service. Penny is PetScripts’ virtual assistant and it’s her job to help customers of the online prescription matching service by answering questions and giving guidance about prescription medicines.
Penny is constantly learning and improving, and if she can’t answer a question she gets a real person to answer, and she then learns that answer for next time. “Penny will talk you through how to shop with us, how to get a script from your vet and those sorts of questions,” says Steven Perissinotto, director of PetScripts.
While Penny doesn’t help vets directly, Perissinotto says in the future he expects more AI technology that helps with client bookings and questions to hit the market in Australia. “More and more clients have questions, and less and less do vets have the time to answer those questions,” he says.
“And then there’s the booking side of things. Mrs Jones rings up and says, ‘My seven-year-old Staffie needs to see a vet—can I come in at 3pm?’. At the moment, a human says, ‘I’m sorry, we don’t have anyone available at 3pm. Does 4.30pm work for you?’ In the future, we can use AI for that.”
Rather than putting people out of a job, Perissinotto says AI has the potential to help practices combat chronic staff shortages. “One of our biggest challenges is that we can’t get enough good vets,” he says. “AI can help. It’s not putting anyone out of a job; it’s doing a job we can’t get anyone to do.”
More data please
But there’s still a way to go before AI becomes commonplace in Aussie vet practices. At the moment it’s used in farming applications like robotic milking, but elsewhere one of the biggest issues is generating big enough data sets for individual species.
“If you look at dogs and we start training an artificial intelligence app with 100,000 cases, that will probably give you a very good idea of what’s happening with really common breeds like labradors,” says Dawson.
“But for really unusual breeds where there’s only a handful in each country, even 100,000 cases aren’t going to give you very much data about those breeds and it won’t be statistically valid. You have to have huge databases to start covering the full range of pets that people have.”
That said, AI technology moves quickly, and experts believe things like veterinary image recognition may be easily accessible—and affordable—in as little as five years. “What most of the applications will be doing is tackling individual problems and acquiring data on that problem,” says Dr Khatkar. “Veterinary imaging can be automated, and this technology will definitely be available soon.”
In the meantime, what should vets do to prepare for AI? “Be ready,” says Perissinotto. “One day, what you imagined may cost you a million bucks will cost you $59 a month.”