How AI will help better anticipate future pandemics 🦠

In a perspective article published in Nature, researchers from Africa, the Americas, Asia, Australia, and Europe describe for the first time how artificial intelligence (AI) can revolutionize infectious disease research and improve pandemic preparedness.

Over the next five years, integrating AI into national response systems could help predict the origin and trajectory of outbreaks, thereby saving more lives. An international group of researchers calls for better collaboration between academic, governmental, and industrial sectors to ensure the safe, responsible, and ethical use of AI in infectious disease research.


A study published on February 20, 2025, in Nature describes for the first time how advances in AI can accelerate progress in infectious disease research and outbreak response.

The study is the result of a partnership between researchers from the University of Oxford, the University of Copenhagen, and the Institut Pasteur, in collaboration with peers from academic, industrial, and political spheres across Africa, the Americas, Asia, Australia, and Europe, who advocate for collaboration and transparency in both datasets and AI models.

Until now, medical applications of AI have primarily focused on individual patient care, such as improving clinical diagnostics, precision medicine, or supporting treatment decisions.

This study, however, focuses on using AI for population health. The research shows that recent advances in AI methodologies are making them increasingly effective, even with limited data volumes. AI's improved ability to process noisy and sparse data opens new opportunities for tools designed to enhance health in both high- and low-income countries.

Lead author Professor Moritz Kraemer from the Pandemic Sciences Institute at the University of Oxford states: "Over the next five years, AI could revolutionize pandemic preparedness. By leveraging terabytes of regularly collected climate and socioeconomic data, it will help us better predict where outbreaks will emerge and their trajectory. It could also allow us to anticipate their impact on individual patients by studying interactions between the immune system and emerging pathogens. Combined and integrated into national response systems, these advances could save lives and better prepare the world for future pandemic threats."

The research identified the following opportunities for AI in pandemic preparedness:
- Promising advances in improving current disease spread models, enabling more robust, accurate, and realistic modeling.
- Progress in identifying high-transmission areas, facilitating the efficient allocation of limited healthcare resources.
- Potential to enhance genetic disease surveillance data, accelerating vaccine development and the identification of new variants.
- Potential assistance in identifying the properties of new pathogens, predicting their traits, and determining the likelihood of cross-species transmission.
- Anticipation of emerging variants of circulating pathogens, such as SARS-CoV-2 and influenza viruses, and the treatments and vaccines most likely to mitigate their impact.
- Potential for AI-assisted integration of population-scale data with individual-level sources, including heart rate monitors and wearable step counters, to better detect and monitor outbreaks.
- AI-powered creation of a new interface between highly technical science and healthcare professionals with limited training, strengthening the capacity of institutions most in need of these tools.

However, AI advances will not have the same impact on all aspects of pandemic preparedness and response. For example, while protein language models show great promise in accelerating the understanding of how viral mutations influence disease spread and severity, fundamental model improvements may only offer modest gains over existing approaches to modeling pathogen transmission rates.

The researchers caution against relying solely on AI to address infectious disease challenges but suggest that integrating human feedback into AI modeling workflows could help overcome current limitations.

The authors are particularly concerned about the quality and representativeness of training data, restricted accessibility of AI models for the broader community, and potential risks associated with deploying black-box models for decision-making.

Study author Professor Eric Topol, founder and director of the Scripps Research Translational Institute, adds: "AI offers tremendous transformative potential for mitigating pandemics, but it depends on extensive international collaboration and comprehensive, continuous surveillance data."

Co-lead author Samir Bhatt from the University of Copenhagen adds: "Infectious disease outbreaks remain a constant threat, but AI provides policymakers with powerful new tools to make informed decisions about when and how to intervene."

"AI offers many opportunities to improve epidemic and pandemic response. In terms of research, the coming years—where we explore how best to use these new technologies—should be particularly exciting."

The authors propose strict evaluation criteria for AI models and advocate for close collaboration between governmental, societal, industrial, and academic sectors to sustainably and practically develop models that improve human health.