Researchers Will Deploy AI to Better Understand Coronavirus

In the months since the novel coronavirus emerged in Wuhan, China, last December, almost 2,000 research papers have been published on the health effects of the new virus, possible treatments, and the dynamics of the resulting pandemic.

This outpouring of research is a testament to the speed with which science can tackle big problems. But it also presents a headache for anyone wanting to stay up to date with the literature, or hoping to mine it for insight about the virus, its behavior, or possible treatments.

Naturally, some believe that artificial intelligence may help. Monday, the White House announced a project in collaboration with tech companies and academics to make a huge amount of coronavirus research accessible to AI researchers and their algorithms for the first time.

The effort will ask AI to mine through the avalanche of research to answer questions that could help medical and public health experts. By cross-referencing papers and searching for patterns, AI algorithms might help discover new possible treatments or factors that make the virus worse for some patients.

Machine learning has huge potential to help wrangle and draw insights from scientific research. But some experts say the approach is at an early stage and is unlikely to help address the current crisis, where the US suffers from more basic needs, like a shortage of test kits.

Microsoft Research, the National Library of Medicine, and the Allen Institute for AI (Ai2), gathered and prepared over 29,000 papers related to the new virus and the wider coronavirus family, 13,000 of them processed so that computers can read the underlying data, plus information about the authors and their affiliations. Kaggle, a platform that runs data science competitions, is creating challenges around 10 key questions related to the coronavirus. These range from questions about risk factors and treatments that do not involve drugs, to the genetic properties of the virus and efforts to develop vaccines. The project also involves the Chan Zuckerberg Initiative and the Center for Security and Emerging Technology at Georgetown University.

“I think the initiative is definitely worthwhile,” says Giovanni Colavizza, an assistant professor at the University of Amsterdam and a visiting researcher at the Alan Turing Institute. “Whether interesting findings will come from these initiatives remains to be seen, but this initiative highlights the importance of structured, open, and programmatic access to the scientific literature.”

Mining scientific papers has sometimes proven useful, finding, for example, connections that suggested magnesium might treat migraines. The hope is that AI will accelerate insights into the novel coronavirus by finding more subtle connections across more data.

Despite an occasionally frosty relationship with big tech, the White House has been meeting with tech executives in an effort to find solutions to the coronavirus crisis. “High tech in general has gotten something of a bad rap, but something like this crisis shows how AI can potentially do a world of good,” says Oren Etzioni, CEO of Ai2. “The scientific literature on the coronavirus is growing exponentially.”

John Brownstein, an expert on health bioinformatics at Harvard Medical School, says the effort is worthwhile, and it is good to see so many people trying to help. At the same time, he notes that worthwhile data projects such as Predict, which is designed to predict pandemics, have been starved of funding in recent years. He also says the government should have been prepared in advance for pandemics, citing a lack of testing kits as a big problem. “We’ve had a severe lack of funding and resources,” Brownstein says. “We want to think about the bigger picture.”

After the US and other governments last week called for scientific publishers to open up research on the coronavirus, a number of big publishers said they would offer free access to relevant papers and data. Many scientists support the idea of making research more open and accessible generally.