Yearly, global health experts are faced with a high-stakes decision: Which influenza strains should go into the subsequent seasonal vaccine? The alternative have to be made months prematurely, long before flu season even begins, and it could often feel like a race against the clock. If the chosen strains match those who flow into, the vaccine will likely be highly effective. But when the prediction is off, protection can drop significantly, resulting in (potentially preventable) illness and strain on health care systems.
This challenge became much more familiar to scientists within the years through the Covid-19 pandemic. Think back to the time (and time and time again), when latest variants emerged just as vaccines were being rolled out. Influenza behaves like the same, rowdy cousin, mutating continuously and unpredictably. That makes it hard to remain ahead, and due to this fact harder to design vaccines that remain protective.
To scale back this uncertainty, scientists at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and the MIT Abdul Latif Jameel Clinic for Machine Learning in Health got down to make vaccine selection more accurate and fewer reliant on guesswork. They created an AI system called VaxSeer, designed to predict dominant flu strains and discover probably the most protective vaccine candidates, months ahead of time. The tool uses deep learning models trained on a long time of viral sequences and lab test results to simulate how the flu virus might evolve and the way the vaccines will respond.
Traditional evolution models often analyze the effect of single amino acid mutations independently. “VaxSeer adopts a big protein language model to learn the connection between dominance and the combinatorial effects of mutations,” explains Wenxian Shi, a PhD student in MIT’s Department of Electrical Engineering and Computer Science, researcher at CSAIL, and lead writer of a brand new paper on the work. “Unlike existing protein language models that assume a static distribution of viral variants, we model dynamic dominance shifts, making it higher fitted to rapidly evolving viruses like influenza.”
An open-access report on the study was published today in Nature Medicine.
The long run of flu
VaxSeer has two core prediction engines: one which estimates how likely each viral strain is to spread (dominance), and one other that estimates how effectively a vaccine will neutralize that strain (antigenicity). Together, they produce a predicted coverage rating: a forward-looking measure of how well a given vaccine is more likely to perform against future viruses.
The dimensions of the rating might be from an infinite negative to 0. The closer the rating to 0, the higher the antigenic match of vaccine strains to the circulating viruses. (You’ll be able to imagine it because the negative of some sort of “distance.”)
In a 10-year retrospective study, the researchers evaluated VaxSeer’s recommendations against those made by the World Health Organization (WHO) for 2 major flu subtypes: A/H3N2 and A/H1N1. For A/H3N2, VaxSeer’s selections outperformed the WHO’s in nine out of 10 seasons, based on retrospective empirical coverage scores (a surrogate metric of the vaccine effectiveness, calculated from the observed dominance from past seasons and experimental HI test results). The team used this to judge vaccine selections, because the effectiveness is just available for vaccines actually given to the population.
For A/H1N1, it outperformed or matched the WHO in six out of 10 seasons. In a single notable case, for the 2016 flu season, VaxSeer identified a strain that wasn’t chosen by the WHO until the next yr. The model’s predictions also showed strong correlation with real-world vaccine effectiveness estimates, as reported by the CDC, Canada’s Sentinel Practitioner Surveillance Network, and Europe’s I-MOVE program. VaxSeer’s predicted coverage scores aligned closely with public health data on flu-related illnesses and medical visits prevented by vaccination.
So how exactly does VaxSeer make sense of all these data? Intuitively, the model first estimates how rapidly a viral strain spreads over time using a protein language model, after which determines its dominance by accounting for competition amongst different strains.
Once the model has calculated its insights, they’re plugged right into a mathematical framework based on something called atypical differential equations to simulate viral spread over time. For antigenicity, the system estimates how well a given vaccine strain will perform in a typical lab test called the hemagglutination inhibition assay. This measures how effectively antibodies can inhibit the virus from binding to human red blood cells, which is a widely used proxy for antigenic match/antigenicity.
Outpacing evolution
“By modeling how viruses evolve and the way vaccines interact with them, AI tools like VaxSeer could help health officials make higher, faster decisions — and stay one step ahead within the race between infection and immunity,” says Shi.
VaxSeer currently focuses only on the flu virus’s HA (hemagglutinin) protein,the most important antigen of influenza. Future versions could incorporate other proteins like NA (neuraminidase), and aspects like immune history, manufacturing constraints, or dosage levels. Applying the system to other viruses would also require large, high-quality datasets that track each viral evolution and immune responses — data that aren’t at all times publicly available. The team, nevertheless is currently working on the methods that may predict viral evolution in low-data regimes constructing on relations between viral families
“Given the speed of viral evolution, current therapeutic development often lags behind. VaxSeer is our try and catch up,” says Regina Barzilay, the School of Engineering Distinguished Professor for AI and Health at MIT, AI lead of Jameel Clinic, and CSAIL principal investigator.
“This paper is impressive, but what excites me even perhaps more is the team’s ongoing work on predicting viral evolution in low-data settings,” says Assistant Professor Jon Stokes of the Department of Biochemistry and Biomedical Sciences at McMaster University in Hamilton, Ontario. “The implications go far beyond influenza. Imagine with the ability to anticipate how antibiotic-resistant bacteria or drug-resistant cancers might evolve, each of which might adapt rapidly. This sort of predictive modeling opens up a strong latest way of fascinated by how diseases change, giving us the chance to remain one step ahead and design clinical interventions before escape becomes a significant problem.”
Shi and Barzilay wrote the paper with MIT CSAIL postdoc Jeremy Wohlwend ’16, MEng ’17, PhD ’25 and up to date CSAIL affiliate Menghua Wu ’19, MEng ’20, PhD ’25. Their work was supported, partially, by the U.S. Defense Threat Reduction Agency and MIT Jameel Clinic.