Picture of health: How feline obesity diagnosis may change in the era of AI
By Troy Johnson
Type the social media hashtag “fatcat” into an Instagram search and you will be greeted by more than 3 million photo and video posts featuring felines in varying stages of roly-polyness. One photo depicts a pet owner straining to lift a 38-pound orange marmalade-colored bundle of joy. Another video shows the fabric seat of a lawn chair collapsing beneath the weight of a chonky black-and-white cat.

While the antics of hefty Himalayans may drive social media engagement, carrying extra weight can have dire consequences for our feline friends. According to the Association for Pet Obesity Prevention, approximately 60 percent of the cats in the U.S. are classified as overweight or obese.
“There are a lot of happy fat cats out there,” said Dr. Xu Wang, professor of comparative genomics in the Auburn University College of Veterinary Medicine’s Department of Pathobiology, “but many obese cats have reduced lifespans due to diabetes and other health complications directly linked to obesity.”
Wang and an interdisciplinary team of colleagues from the College of Veterinary Medicine, Auburn’s College of Sciences & Mathematics, and Nestlé Purina PetCare have taken a step toward unlocking possibilities to expand feline healthcare and research and leverage artificial intelligence in new ways. Their article, “Inter-evaluator bias and applicability of feline body condition score from visual assessment,” newly published in Frontiers in Veterinary Science, explores the trusted Body Condition Score (BCS) diagnostic tool used by veterinarians to evaluate body fat in cats.
Developed in 1997 by Nestlé Purina PetCare Company, BCS is a visual and hands-on assessment that rates cats on a 9-point scale, with a score of one representing emaciation and nine representing severe obesity. A score of four or five is considered ideal for cats, while a six indicates a cat may be slightly overweight.
“BCS represents the current gold standard tool in veterinary practice,” Wang said.
The research team decided to test whether BCS could be accurately scored solely from photographs rather than the usual clinic-based approach. A series of online-sourced cat images was administered as a quiz to nine trained evaluators. To validate the results, top and side images of 38 cats were gathered during routine wellness exams. A clinician gathered BCS data on each of the cats through “palpation,” or hands-on examination for thickness of the fat layer. Nine evaluators later assessed BCS for the same cats visually, using only photographs.

The results? There was minimal variance between the BCS exams conducted by palpation and assessments rendered through review of images. “The bias was very low,” Wang said. “The mean was less than one point. The bias was lower than we expected.” These findings mean that the BCS assigned to a patient using only visual cues was extremely similar to those using both visual and palpation assessments in the clinic. They also confirmed that different trained clinicians would give the same feline patient a similar BCS, which is important for providing consistent healthcare.
The expertise of Auburn’s research team encompasses feline medicine, comparative genomics, metabolic diseases, internal and preventative veterinary medicine, neurodegenerative diseases, and data science and machine learning. In addition to Wang, Auburn University’s research co-authors include Drs. Emily Graff, Christopher Lea, Diane Delmain, Erin Chamorro, and Doug Martin from the College of Veterinary Medicine, and Dr. Jingyi Zheng from the College of Sciences & Mathematics. Drs. Lea, Delmain, and Chamorro are clinicians at Auburn University Veterinary Clinic, which is a feline friendly primary-care veterinary practice. Delmain is feline diplomate of the American Board of Veterinary Practitioners (ABVP). Dr. Johnny Li, a senior research scientist at Nestlé Purina PetCare and an affiliate professor of Auburn’s College of Veterinary Medicine, collaborated with the research team.
Other contributors included Dr. Emily Brinker at Tufts Cummings School of Veterinary Medicine, a former resident and Ph.D. student with Dr. Graff, as well as Xiaolei Ma and Yue Zhang, Ph.D. students with Dr. Wang, and Auburn DVM program graduates Kenzii Kittell, Mackenzie Hicks, Casey Pfister and Heather Hamilton. This project is supported by the non-profit EveryCat Health Foundation, a leader in advancing feline medicine and improving the lives of cats worldwide through the support of groundbreaking research.
“These findings on feline health, diet, and weight management highlight the vital role of research we facilitate and our donors are proud to support — as the world’s only granting organization dedicated exclusively to advancing cat health through science on a global scale,” EveryCat Health Foundation President and CEO Jackie Ott-Jaakola said.
Advancing telemedicine and AI
The implications for the research are far-reaching in a world where access to portable technology often outpaces access to convenient veterinary care.
“Cats need routine wellness visits just like any of our veterinary patients,” said Graff, first and co-senior author of the study. “Unfortunately, the logistics of bringing a healthy cat to the clinic can be challenging for owners, which creates a huge gap in our ability to provide good feline healthcare. We need better options to reach the larger cat community.”
Among the options is data processing capability that exists in cell phones and smart home devices.
“One of the potential next steps involves artificial intelligence,” Wang said.
Building a larger database of feline images and combining that with clinical BCS evaluation data could put a reliable diagnostic tool in the palm of pet owners’ hands. This method could power AI-enabled mobile apps that allow pet owners to snap photos and receive a BCS assessment instantly. It could also open additional telemedicine possibilities for pet owners who may lack access to transportation or who live in communities with limited veterinary care options.

“Pet owners could use this as a starting point, and then seek further advice,” Wang said. Pet owners may also be able to conduct regular assessments to help with weight loss program compliance.
The technique of BCS assessment via image analysis also offers greater flexibility for clinicians and reduces anxiety for pet owners as well as their cats, who may experience stress on the examination table.
Beyond telemedicine and mobile app-enabled management of feline diet and nutrition and long-term health monitoring, the visual BCS method also creates possibilities for supporting veterinary education, offering a bank of visual examples for first-year DVM students to practice BCS assessments before working with live animals.