How Your Dog or Cat Could Help Speed Up Your Health Tests
Summary
The Mayo Clinic, Quest Diagnostics and others are exploring AI tools that build on work with animal data to train diagnostic algorithms for humansImagine getting medical-test results within minutes or seconds, before you leave the doctor’s office, for even the most complicated diagnoses.
Steps toward that goal are happening in an unlikely place: The veterinarian’s office. Researchers are using results from your pets’ tests for things like hookworms and other parasites to train artificial intelligence tools that not only speed up Fido’s results, but are informing human diagnostics, too.
The Mayo Clinic this year began using diagnostic algorithms developed by a Salt Lake City-area startup called Techcyte that uses its work with animal data to inform its AI development. Mayo’s large-scale laboratory, which processes samples for its own patients and other health systems, used to have a two-week backlog of parasite samples. Now it can finish the 200 that arrive each day in 24 hours.
Big testing companies such as Quest Diagnostics will start using Techcyte algorithms next year for fecal testing for parasites. And Techcyte says it is trying to win clearance from U.S. health regulators to allow smaller sites such as doctor’s offices to use algorithms in tests that would give patients results before their appointment ends.
Techcyte is also working with Mayo to develop AI-based cancer-detection tests. The AI can give pathologists and physicians more data to determine whether a patient has cancer, including doctors outside academic medical centers.
“We can produce results faster, but also the algorithm is unfailing," said Dr. Bobbi Pritt, interim chair of Mayo’s Department of Laboratory Medicine and Pathology. “It doesn’t suffer from human error, and so it doesn’t get distracted. It doesn’t get fatigued."
Veterinarians are providing the groundwork as they increasingly use artificial intelligence in caring for dogs, cats and other pets or livestock. Among those selling AI-based diagnostic tests for animals are Antech and Idexx Laboratories.
Drugmaker Zoetis, which manufactures pharmaceuticals for farm animals and pets, worked with Techcyte to develop a diagnostic device called Vetscan Imagyst. Zoetis says more than 3,000 veterinary practices in the U.S. use the device today for AI-based fecal, blood, dermatologic and urine tests.
The Imagyst digitizes the process of studying samples with a microscope while giving reliable results that come in a matter of minutes instead of having to wait hours or several days from lab technicians.
“The pain in this market is the same whether you’re a vet, a human lab, or an environmental lab," said Ben Cahoon, chief executive of Techcyte. “Looking at a microscope is a very difficult and tedious job, meaning you need someone who’s an expert. They are looking at a microscope literally for eight hours a day."
The Imagyst looks like a traditional microscope but works as a digital scanner. A vet tech prepares a sample on a glass slide—that step can still take 10 minutes—and puts it on the machine, which uploads the image to Techcyte’s cloud. The algorithm goes to work reading the image and generates results in a few minutes.
The most popular test, using fecal matter, is among the most common tasks performed by veterinarians; experts recommend that pet owners get their pets checked for parasites at least once a year. A blood smear test, which can count red and white blood cells, and platelets, gives results in less than four minutes. Vets normally send blood samples out to labs.
To design its fecal test, Zoetis collected roughly 50,000 samples from veterinarians and laboratories around the world, many containing more than 100 parasites like hookworms or tapeworms. After its scientists marked the parasites, the images were fed to Techcyte’s algorithm. The findings were validated against the findings of human scientists.
The Imagyst accuracy was comparable to findings of clinical specialists, according to Zoetis-funded studies published in the journal Parasites & Vectors beginning in 2020.
Joseph D’Abbraccio, a veterinarian about a two-hour drive from New York City in the Hudson Valley, runs his Imagyst devices about 20 times a day.
D’Abbraccio says his staff read the results on computers while caring for animals instead of waiting around in the lab. The fast turnaround on test results means pets don’t need to be hospitalized meanwhile, which can cost hundreds of dollars. “I get a faster clinical judgment on the patient, which then in the end decreases my client’s bills, because the animal’s not waiting here for days for tests. I get results before lunch."
Animal health wasn’t part of Techcyte’s initial business plan when it started in 2013 and focused on human health. Cahoon joined as chief executive three years later, when the company still hadn’t launched a product. Attending a medical conference, he learned the animal health industry faced a backlog of studying stool samples—a “fecal tsunami," he says he heard.
He thought developing tests for animals and the environment would be faster because they have fewer regulatory barriers. At trade shows, Cahoon wore a full-size poop costume pitching the company. Zoetis started working with Techcyte in 2019, and two years later became an investor.
Meanwhile, Techcyte continued researching AI for people, building on its learnings from animal health. For example, the framework for training the animal AI tests—evaluating different values and settings—was used for human tests as well, since they’re both microscopy.
The testing company ARUP Laboratories in 2019 began using a Techcyte algorithm for a human fecal test. Techcyte rolled out bacteriology tests last year at health systems like Mayo.
Techcyte’s research was getting noticed, including by Mayo’s Pritt. A 2020 peer-reviewed study in the Journal of Clinical Microbiology by researchers at Techcyte and ARUP found that the company’s tools were 98.88% consistent with conventional microscopy and more sensitive than humans.
Pritt said the animal-health success added credibility to Techcyte. “Looking for parasites and stool samples is very similar for animals as it is for humans," she said.
At Mayo’s laboratory, the AI helped to slightly increase the positivity rate of samples to 5%, Pritt said. The AI also sped up the review of negative samples to 30 seconds, instead of five minutes done by a staff member. Because most slides come back negative, workers no longer spend most of their time screening slides without parasites.
Technologists at Mayo still confirm the positives identified by AI.
Mayo is also working with Techcyte to develop a digital pathology program and AI-based tests on tissue samples instead of just liquids and cells, including tests for cancer, said Maneesh Goyal, chief operating officer of Mayo Clinic Platform, which invests in startup firms including Techcyte.
Mayo is in the process of digitizing some 25 million slides of tissue samples collected from patients, and for all patients in the future. These images will train the algorithms, which may be ready for use in care next year, Goyal said.
Write to Jared S. Hopkins at jared.hopkins@wsj.com