Researchers at Washington State University have developed a machine learning tool with the potential to head off pandemics. The model, designed for identifying which animal species might be capable of harboring and spreading viruses dangerous to humans, has a specific focus on orthopoxviruses, a group that includes the smallpox and mpox viruses. According to a recent WSU press release, the findings of this research could help scientists to preemptively target zoonotic threats.
Under the guidance of Stephanie Seifert, an assistant professor at WSU's Paul G. Allen School for Global Health, the team's work could direct efforts in staving off future viral outbreaks. "Nearly three-quarters of emerging viruses that infect humans come from animals," Seifert told WSU News. Exploring the genetic makeup of viruses as well as host characteristics, the model pinpoints potential hotbeds for outbreaks, especially in areas like Southeast Asia, equatorial Africa, and the Amazon – locations with not only a high population of potential hosts but a lapse in smallpox vaccination that provides incidental protection against other orthopoxviruses.
The machine learning tool goes beyond traditional methods, which tend to focus largely on the ecological traits of organisms. It also improves upon previous models by including virus genetics in its assessment of which species are most likely to host and disseminate diseases. "Our model improves the accuracy of host predictions and provides a clearer picture of how viruses may spread across species," said Pilar Fernandez, an assistant professor in the Allen School, who contributed to the research, as per WSU News.
The system has identified key animal families as potential hosts for mpox, including rodents, cats and canids (such as dogs and related species), skunks, mustelids (like weasels and otters), and raccoons. However, it has also correctly ruled out certain species like rats, known to be resistant to mpox infection. The predictive model stands as a more efficient alternative to traditional field sampling which can be time and resource-intensive. Seifert highlighted the pragmatic application of the tool: "If we can use these machine learning models to help us prioritize sampling efforts, then that's going to be really beneficial in identifying where these viruses are coming from and in understanding the risks they pose," WSU stated.
