Sapienza University researchers develop AI to predict zoonotic virus jumps

Researchers at Sapienza University of Rome have developed an artificial intelligence model designed to identify potential links between viruses and animal species before they trigger global pandemics. The interdisciplinary study, conducted by the departments of Physics and Biology and Biotechnology “Charles Darwin,” was published in Nature Communications.

Because less than 1% of mammalian zoonotic viruses are currently known to science, existing databases are incomplete. This creates a statistical challenge where models may mistake a lack of data for the actual absence of a virus. To address this, the team developed a technique called Dynamic Positive-Unlabeled (DPU) learning.

This machine learning approach distinguishes between confirmed virus-host associations and mere gaps in scientific knowledge. By integrating pathogen evolution, geographic distribution, and mammalian biological characteristics, the model aims to help health authorities prioritize surveillance resources and map biological threats more effectively.