Predictive Modeling of Global SARS-CoV-2 Infection Risk in Animals: Unveiling Potential Reservoirs and Informing...

Area of Science:

• Veterinary Medicine
• Epidemiology
• Machine Learning

Background:

• Reports of SARS-CoV-2 (the virus that causes COVID-19) in animals raise concerns about natural reservoirs.
• Global distribution and drivers of animal infection risk are not well understood.

Purpose of the Study:

• To estimate the global probability of SARS-CoV-2 infections in animals using machine learning.
• To map current infection risk and project future risk in areas with limited data.

Main Methods:

• Extensive data mining from diverse sources.
• Development and evaluation of three machine learning models.
• Risk mapping in well-documented and data-sparse regions.

Main Results:

• High-risk areas identified in Europe, the US, southern Brazil, and Asia.
• Overlaps between high-risk zones and distributions of white-tailed deer, American mink, and Asian small-clawed otters.
• Anthropogenic factors (accessibility, population density, COVID-19 mortality) were more predictive than biophysical factors.

Conclusions:

• Machine learning models can estimate and map global SARS-CoV-2 animal infection risk.
• Human activities significantly influence the risk of SARS-CoV-2 transmission to animals.
• Findings support integrating public and animal health policies for effective disease management.