Viruses like SARS-CoV-2 constantly mutate to develop new ways of escaping our immune defenses. This process creates an ongoing “arms race” between viruses and our immune systems, as each tries to outsmart the other. Vaccines have been one of humanity’s most powerful tools in this fight, helping to eradicate diseases like smallpox and saving millions of lives each year. However, some viruses mutate so quickly that it’s challenging to stay ahead. To tackle this, scientists are developing advanced tools to predict how viruses might evolve and how to prepare for future threats before they arise.
AI and Predictive Models of Viral Evolution for SARS-CoV-2
In March 2022, a multidisciplinary team of scientists at Harvard Medical School launched a groundbreaking project aimed at using artificial intelligence (AI) to forecast how SARS-CoV-2 might evolve. By analyzing patterns in how the virus mutates, Harvard’s AI models can predict which changes might allow the virus to evade vaccine-induced or natural immunity. This helps address one of the biggest challenges in pandemic preparedness: staying ahead of the virus.
The tools developed through this project have two key benefits:
Improved monitoring of new variants: The AI models allow scientists to quickly identify mutations in new strains that could help the virus evade vaccines or immunity from previous infections.
Better vaccine design: By understanding how the virus is likely to change, scientists can create vaccines that provide broad, long-lasting protection against both current and future variants.
Illustration above by Matt Chinworth
Harvard’s Contribution to Preparing for H5N1 (Avian Flu)
In August 2024, Harvard expanded its work to include H5N1, a strain of avian influenza (bird flu) with the potential to cause a future pandemic. This effort is part of CEPI’s “low regret response” strategy to bolster preparedness against a potential escalation in H5N1 risk. Harvard is working closely with other CEPI’s awardees, including the Houston Methodist Research Institute, to accelerate vaccine development for H5N1.
The Harvard team’s AI models are playing a vital role in this work by:
Designing vaccines with broad protection: The goal is to create vaccines that work against many variations of H5N1, even if the virus mutates in unexpected ways.
Making vaccines future-proof: The AI helps ensure that vaccines remain effective over time, even as the virus continues to evolve.
Building a Safer Future Through Predictive Modelling
By combining cutting-edge AI with insights from biology, Harvard’s work is helping to build a safer future, where we can respond to viral threats more quickly and effectively. This approach not only strengthens our ability to fight current threats but also lays the groundwork for tackling future pandemics before they start.
