AI Model Enhances Prediction of Animal Hosts for Virus Spillover Risks

Researchers at Washington State University have developed a machine learning model that could help predict and prevent pandemics by identifying animal species likely to harbor viruses capable of infecting humans. The model, designed to analyze both host characteristics and virus genetics, focuses on orthopoxviruses, including the viruses responsible for smallpox and mpox.

The study, published in Communications Biology, highlights how this tool could assist scientists in anticipating zoonotic threats and adapting the model for other viruses. “Nearly three-quarters of emerging viruses that infect humans come from animals,” said Stephanie Seifert, senior author on the study. “If we can better predict which species pose the greatest risk, we can take proactive measures to prevent pandemics.” 

The model identified regions such as Southeast Asia, equatorial Africa, and the Amazon as hotspots for orthopoxvirus outbreaks. These areas have high concentrations of potential animal hosts and low smallpox vaccination rates. Although the smallpox vaccine offers cross-protection against orthopoxviruses, vaccination efforts ceased after smallpox eradication in 1980.

Several animal families were flagged as potential mpox hosts, including rodents, cats, canids (e.g., dogs), skunks, mustelids (e.g., weasels), and raccoons. Notably, the model excluded rats, aligning with laboratory findings showing their resistance to mpox infection. Katie Tseng, the study’s lead author, noted that the model outperformed previous approaches. “While we used the model specifically for orthopoxviruses, we can also go in a lot of different directions and start fine tuning this model for other viruses.”

Unlike earlier models that relied solely on host ecological traits, this new tool incorporates virus genetics. Pilar Fernandez, co-senior author, explained that this dual focus improves prediction accuracy and clarifies how viruses spread across species. “Previous models were more based on the characteristics of the host, but we wanted to add the other side of the story, the characteristics of the viruses. Our model improves the accuracy of host predictions and provides a clearer picture of how viruses may spread across species.”

By streamlining wildlife surveillance efforts, the model offers a practical approach to identifying virus reservoirs and prioritizing sampling in biodiverse regions like Central Africa. This could play a crucial role in mitigating future spillover events.