Dogs can signal illness in subtle ways through their social behaviour, finds new research from the Ontario Veterinary College (OVC).

Michael Brunt smiles at the camera with curly brown hair and black eye glasses wearing a plaid collared shirt under a light brown blazer.
Dr. Michael Brunt

In a study recently published in the Journal of the American Association for Laboratory Animal Science, OVC and international researchers found dogs that are feeling unwell interacted less often with other dogs.

“The behaviour of an animal is one of the best ways to assess their welfare,” said lead author Dr. Michael Brunt, postdoctoral fellow in the Department of Population Medicine and researcher in The Campbell Centre for the Study of Animal Welfare.

The study involved observing the behaviour of 12 mature, female beagles in a controlled environment.

The dogs were fed a mix of three diets, some contaminated with Fusarium mycotoxin, a toxin produced by mould and often found in cereal-based pet food that can cause a variety of ailments.

Researchers then individually released the dogs into the centre aisle of a housing room for four minutes per day, observing their interactions with familiar dogs in adjacent kennels.

They found the total number of interactions, orientation, and attempted physical contact with other dogs were less frequent among those that had consumed mycotoxins.

Study an opportunity for develop AI, machine learning

“There is no one who knows their dog better than the humans who live with them,” Brunt said. “You know when your dog isn’t itself.” A pet owner may not be able to articulate the exact symptoms a dog is experiencing, but they can sense changes in their behaviour.

“With this study,” Brunt said, “we were able to pinpoint what some of those changes are, such as wanting to interact with other dogs less, or not seeking out social interactions with other dogs.”

The changes were subtle, he said, but significant.

“This is an opportunity to look at precision technologies and develop machine learning algorithms that could pick up those subtle changes in behaviour in a lab setting,” he explained.

“It could be a way to identify subclinical illness before it compromises their welfare,” he added.

Brunt stressed this study focused on one dog breed within a research environment; he would like to see further studies conducted with other breeds in a variety of environments.

This research was supported by the Natural Sciences and Engineering Research Council of Canada.


Dr. Michael Brunt