Cookies on VetMed Resource

Like most websites we use cookies. This is to ensure that we give you the best experience possible.

 

Continuing to use www.cabi.org  means you agree to our use of cookies. If you would like to, you can learn more about the cookies we use.

VetMed Resource

Veterinary information to support practice, based on evidence and continuing education

Sign up to receive our Veterinary & Animal Sciences e-newsletter, book alerts and offers direct to your inbox.

News Article

Machine learning improves early identification of chronic kidney disease in cats


The use of models based on machine learning can support veterinary decision making

Antech Diagnostics, part of Mars Petcare, has developed a diagnostic tool using artificial intelligence and veterinary records. The resulting biomarker model can predict chronic kidney disease (CKD) in cats up to two years before traditional diagnosis.

The RenalTech™ tool, available to veterinarians in the U.S., was developed from a research project led by Richard Bradley at WALTHAM Centre for Pet Nutrition, published in the Journal of Veterinary Internal Medicine. The project involved mining the veterinary records of more than 150,000 cats that visited BANFIELD® Pet Hospitals in partnership with the Mars Advanced Research Institute (MARI) and Process Integration and Predictive Analytics (PIPA LLC).

CKD is a complex disease that historically has been difficult to diagnose. The current methods are only able to confirm the disease once significant and irreversible kidney damage has occurred, which makes any intervention and treatment challenging. The new diagnostic tool uses a biomarker of six common feline health measurements (creatinine, blood urea nitrogen, white blood cell count, urine specific gravity, urine protein, urine pH–along with approximate age) to predict CKD before traditional diagnosis.

To develop the diagnostic tool, historic health records spanning 20 years from 750,000 veterinary hospital visits for over 150,000 cats were examined to hunt for changes that were characteristic of the pets that were known to go on to develop CKD, when compared to those who remained healthy. This approach allows for these predictive patterns to be tested to determine how accurate the diagnosis is. For the RenalTech™ tool, they found the accuracy was greater than 95 percent.

“When we looked at the historic data from thousands of pets, it was clear that the data had a story to tell,” said Richard Bradley, author and Data Science Technical Lead at WALTHAM. “There were subtle changes in several of the blood and urine parameters long before disease was diagnosed, but they were different from pet to pet. Machine learning allowed us to imprint all the subtleties of the changes in a computer algorithm, which was then able to spot the small abnormalities and make a robust prediction.”

“This is a paradigm shifting moment for veterinary medicine,” said Jonathan Elliott, Vice Principal for Research and Innovation at the Royal Veterinary College and partner in the development of the tool. “The ability to use artificial intelligence on data collected in practice as part of routine health screens to address chronic kidney disease well before the disease becomes clinically apparent gives us an opportunity to leverage best practice medicine for cats not previously available to veterinarians. It’s also exciting that we were able to move so quickly from publication in the Journal of Veterinary Internal Medicine to making the tool available in practice.”

Article: Bradley, R., Tagkopoulos, I., Kim, M., Kokkinos, Y., Panagiotakos, T., Kennedy, J., De Meyer, G., Watson, P., Elliott, J. (2019). Predicting early risk of chronic kidney disease in cats using routine clinical laboratory tests and machine learning. Journal of Veterinary Internal Medicine, Early View, online 26 September 2019, doi: 10.1111/jvim.15623

Article details

  • Date
  • 05 November 2019
  • Source
  • WALTHAM Centre for Pet Nutrition