Scientists use AI to identify animal populations most likely to spread diseases to humans

Avian flu, mad cow disease, hantavirus, black plague and other notorious ailments originated with animals and made the jump to humans.

Now scientists at Washington State University have built a machine learning model to examine multiple indicators that increase the odds of a disease making that leap, including the ecological characteristics of the host animals, virus genetics and the animals’ overlap with humans.

The team at the WSU College of Veterinary Medicine’s Paul G. Allen School for Global Health is focused on a specific class of zoonotic diseases called orthopoxviruses — which includes the viruses that cause smallpox and mpox, or monkey pox.

“Nearly three-quarters of emerging viruses that infect humans come from animals,” Stephanie Seifert, an expert in viral emergence and cross species transmission who helped lead the research, said in a press release. “If we can better predict which species pose the greatest risk, we can take proactive measures to prevent pandemics.”

Earlier work predicting potential orthopoxviruses focused on animal traits including habitat and diet as well as how they behaved in the environment. The new work added essential information on the genetic make up the viruses.

“Our model improves the accuracy of host predictions and provides a clearer picture of how viruses may spread across species,” said Pilar Fernandez, a disease ecologist who partnered with Seifert on the research.

The tool highlighted potential hotspots for orthopoxvirus outbreaks that include Southeast Asia, equatorial Africa and the Amazon — areas that intersect with human populations with low vaccination rates for smallpox. The model pointed to possible host species such as rodents, cats, dogs and related species, skunks, weasels and raccoons — but not rats, which research shows are resistant to mpox.

The model can be adjusted to search for other sorts of zoonotic diseases.

The researchers have published a study on their model in the journal Communications Biology. The team included members of from the Viral Emergence Research Institute who are located at the University of Oklahoma, University College London and Yale University.