This year’s flu season is shaping up to be a nasty one, so far sending 8,990 Americans to the hospital and killing 30 children. And with this year’s vaccine estimated to be only about 30 percent effective at preventing influenza, researchers are scrambling to find better ways to protect people from flu in years to come.
Artificial intelligence — and machine learning in particular — could hold some solutions.
There are two key ways that AI is being used in flu research: to provide public health officials with better forecasts of flu season and to help scientists develop more effective vaccines — possibly even to develop a universal flu vaccine that would offer protection against all strains of influenza.
Forecasting the flu
Forecasts help researchers track flu outbreaks, avert vaccine shortages, and provide crucial updates for the public before and during each flu season — akin to the way meteorologists provide weather forecasts, said Dr. Roni Rosenfeld, a machine learning expert at Carnegie Mellon University in Pittsburgh.
Rosenfeld and his Carnegie Mellon colleagues use machine learning to comb through historical data on how the flu has spread in past years in various places across the country.
“We make an underlying assumption that this year is going to be, in some senses, similar to a past year,” Rosenfeld said. But he wants to drill down into the data to provide more detailed assessments of the flu’s impact on individual counties and cities — and for good reason.
“When we track and forecast a flu in regions like the Southeast or New England, there’s not a single epidemic going on,” Rosenfeld said. “At the local level, the flu hits different cities and counties in different ways and at different times. They could really use a much more specific and customized assessment of what’s going on.”
Developing better vaccines
As Rosenfeld and other researchers are using AI to improve flu forecasting, others are using it for the tricky task of developing new, more effective flu vaccines.
Every year the Centers for Disease Control and Prevention (CDC) characterizes about 2,000 influenza viruses. The agency uses this information to pinpoint which specific flu viruses are likely to circulate in the population six months down the road — so that other researchers can develop vaccines that target these strains.
This is complicated because flu viruses mutate rapidly, which means the prevailing strain of flu may morph in a way that renders vaccines less than fully effective at preventing illness and controlling symptoms of people who catch flu despite getting vaccinated.
AI could take some of the guesswork out of developing flu vaccines, helping to ensure their effectiveness.
In 2011, Dr. Richard Webby, a virologist at St. Jude Children’s Research Hospital in Memphis, Tennessee, collaborated on a first-of-its-kind study that used machine learning to examine the specific mutations in flu viruses believed to have led swine flu to jump from pigs to humans — causing a 2009 pandemic that the CDC estimates killed 151,700 to 575,400 people around the world.
And machine learning is showing its value as a tool for studying ordinary seasonal flu viruses as well as the ones that cause pandemics. Scientists are now using it to identify viruses’ so-called antigenic properties — the viruses' molecular structures and how they trigger the body’s immune system.
“The most important piece of information that we have is antigenic information on these viruses that are circulating,” Webby said. Scientists are trying to “train machine learning algorithms to predict what impact various mutations might have on the antigenicity of the virus” — in other words, to make sure scientists are getting the vaccine right.
At least one biotech firm is fully embracing this approach using artificial intelligence.
In October 2017, Boston-based biopharmaceutical company Berg teamed up with French drugmaker Sanofi Pasteur to use AI to study why flu vaccines protect some people well but not others.
Berg CEO Niven Narain says AI could lead to personalized flu vaccines — that is, “different vaccines for different ages or types of populations” or even different geographic locations. As he puts it, “AI is absolutely a game-changer.”