There’s a growing number of AI tools tackling the world’s problems—and pandemics, it appears, is one of the latest to get a digital assist.
A new AI system named EVEscape can predict alterations likely to occur to viruses as they evolve—potentially forecasting the next concerning COVID variants and mutations, as well as changes to other viruses like the flu and HIV.
Developed by researchers at Harvard Medical School and the University of Oxford in 2021, the system was detailed in an Oct. 11 article in Nature. Had it been available at the start of the pandemic, EVEscape would have predicted the most frequent mutations and most concerning variants of COVID, its creators contend.
More than two years after the model’s development, its creators say they have amassed a wealth of data to show that such a system “can make surprisingly accurate predictions,” study co-lead authors Sarah Gurev, a PhD student at the Massachusetts Institute of Technology; Nikki Thadani, a postdoctoral research fellow at Harvard Medical School; and Pascal Notin, a PhD candidate at Oxford, tell Fortune.
A new digital tool that plays offense and defense
Researchers first developed EVE—short for evolutionary model of variant effect—in 2020, to predict whether genetic mutations in humans were disease-causing or benign. They soon realized that the same technology might be applied to viruses like COVID, and dubbed the spin-off EVEscape.
EVEscape is currently being used to evaluate up-and-coming COVID variants, assessing which might be the most dangerous. The tool assigns a score to each variant reported to a global database of COVID sequences. The higher the score, the greater the chance of that variant evading immunity from prior infection and vaccination.
It’s no small task. Thousands of new COVID strains are emerging each month, the trio says. The group produces a biweekly report it publishes online and shares with the World Health Organization, among others, to assist in pandemic planning.
In its latest report, EVEscape flagged a number of variants as most concerning, including:
BA.2.86.1—a child variant of the so-called “Pirola” strain, which contained many more mutations than normal.
XBB.1.5—the so-called “Kraken” variant that surged globally earlier this year.
DV.7.1—a distant relative of the original Omicron. (Most Omicron spawns today are descendants of BA.2, the so-called “stealth Omicron.”)Eventually, such data could aid in the tailoring of vaccines and the creation of therapies like monoclonal antibodies—administered in a hospital setting to high-risk patients—and antivirals like Paxlovid.
Even with tools like EVEscape, it’s impossible to say with complete certainty the exact COVID variants will be dominant six months from now—a fact that doesn’t bode well for vaccines as they currently exist.
But systems like it “have great potential to be used by vaccine manufacturers for the design of variant-proof vaccines,” by pointing out target regions of the virus unlikely to mutate in the future—unlike COVID’s continually morphing spike protein.
EVEscape can also be used to play defense in the war against existing viruses with pandemic potential, like Lassa and Nipah, on the WHO’s list of pathogens with pandemic potential. Both start with non-specific ailments not unlike COVID or the flu, but can easily progress to more severe symptoms and death. The Nipah virus, for instance, is fatal in 40% to 75% of cases. COVID’s case fatality rate was 1% to 4%.
Both viruses are vastly understudied, according to the trio. EVEscape will allow researchers to predict immune escape for “every single mutation to these viruses,” allowing them to assemble a “watch list” for changes with the potential to endanger the few therapies that exist.
{Categories} _Category: Takes,_Category: Inspiration,*ALL*{/Categories}
{URL}https://fortune.com/well/2023/10/13/scientists-use-artifical-intelligence-predict-next-big-covid-variants-pandemic-hiv-flu-lassa-nipah-virus/{/URL}
{Author}Erin Prater{/Author}
{Image}https://content.fortune.com/wp-content/uploads/2023/10/GettyImages-1310117462-e1697165236579.jpg?w=2048{/Image}
{Keywords}Health{/Keywords}
