AI Shows Vaccine Could Protect People of Color Less

One size doesn’t fit all

Key Takeaways

  • Researchers used AI to discover that coronavirus vaccines could be less effective for specific populations.
  • Genetics play a role in how well vaccines work, scientists say. 
  • More research needs to be done using AI to study how different vaccines and drugs affect people of color, observers say.
Closeup of a scientist preparing a liquid sample.
boonchai wedmakawand / Getty Images

Using artificial intelligence, researchers have discovered that some coronavirus vaccines may offer less protection to people of color.

The news comes as several COVID-19 vaccines have demonstrated their effectiveness in recent weeks, raising hopes for an end to the pandemic. The vaccines could be up to 95% effective. But there may be variations in how well they work depending on the patient’s race, according to a recent paper in the journal Cell Systems.

"A person's ancestry influences the origin of their genes, and thus the variants they carry," David Gifford, a professor of biological engineering at MIT and one of the paper’s authors, said in an email interview. "Some of the genes we inherit are central to vaccine processing, and thus the effectiveness of a vaccine may slightly vary depending upon ancestry."

Asian-Americans Get Left Behind

Gifford’s team used advanced machine learning AI methods to examine a type of vaccine similar to Moderna's and Pfizer’s. The researchers discovered that race could play a significant factor in how well vaccines work.

While the team found that white populations were projected to see just 0.04% vaccine inefficacy, Black people were projected to see 1.2% inefficacy (30 times higher than white populations), and Asian-American populations were projected to see 10% inefficacy (250 times higher).

Someone wearing gloves, drawing a COVID-19 vaccine into a needle labeled COVID-19.
skaman306 / Getty Images

Gifford and his students also found a new machine learning-based approach that could improve the vaccines' effectiveness with certain populations. They did this by adding a small number of additional Covid-19 peptides (short chains of amino acids) to a given dose of the vaccine. 

The team’s augmented vaccines use vaccine components that have been observed to cause cellular immune system responses in COVID-19 patients. By adding extra peptides to a particular dose, the team says that preliminary tests show they boosted the vaccine’s effectiveness to nearly 100% in all populations. 

The possible decrease in the effectiveness of vaccines among some populations is a particular concern because people of color have been disproportionately affected by the virus. "To have a vaccine which is less effective for minorities compounds the problem," AI researcher Adrian Zidaritz said in an email interview.

"But at the same time, we should not lose sight of the fact that it is with AI techniques that it was discovered that adding additional peptides to a vaccine dose will improve its effectiveness for the entire population, including people of Black and Asian genetic ancestry. So there is both concern and reason for optimism in this data."

"Some of the genes we inherit are central to vaccine processing, and thus the effectiveness of a vaccine may slightly vary depending upon ancestry."

One Vaccine Doesn’t Fit All

There’s been growing interest in using AI for drug development and identifying drug targets. But AI should also be used to discover how drugs affect distinct populations differently, Anant Madabhushi, a professor at the Case Western Reserve University School of Medicine, said in an email interview.

Madabhushi has produced research showing that in diseases like prostate cancer, there are differences in the disease characteristics between different populations, including Africa-American men versus Caucasian men. Using AI, his team has found differences between breast cancer appearance in South Asian women and white women in North America.

A senior adult getting vaccinated by a home health worker.
fstop123 / Getty Images

"All of this seems to reiterate the point that as we think of drug and vaccine development and response, we have to be mindful of differences across populations and sex," he said. 

Examining a potential drug’s efficacy is not the only way AI is helping with drug development. AI can also be used to help select "the right science to prioritize," Mike Wenger, vice president of patient engagement at TrialScope, a company that sells software for clinical trials, said in an email interview. "It can also be used to look at different types of trials to predict enrollment success rates."

With a coronavirus vaccine rollout expected soon in the U.S., it’s important that everyone get equal access to effective treatments and vaccines. AI is helping to make sure that some Americans don’t get left behind in the race to stop the pandemic.

Was this page helpful?