Caitlin Rivers, an epidemiologist and senior scholar at the Johns Hopkins Center for Health Security, has been thinking about this issue for more than a decade, since working on pandemic-prediction initiatives for the Obama administration. (Yes, better pandemic prediction has been under discussion at least that long.) She laid out details last year in a proposal for a national center for epidemic forecasting, published in Foreign Affairs, with Dylan George of the intelligence-focused venture capital firm In-Q-Tel.
They wrote that pandemic prediction suffers from relying on academics who have to justify their research to grantmakers, and who can’t necessarily step away when public service needs their expertise. The authors proposed giving disease modelers financial support to work out their models in advance of emergencies, and creating formal channels between them and federal decisionmakers who could call on their work as needed—similar to what the National Weather Service already does.
Rivers’ and George’s proposal was read by the right people. Five days after President Joe Biden’s inauguration, the new administration committed to creating a National Center for Epidemic Forecasting and Outbreak Analytics. In March, they designated $500 million in funding for it as part of the American Rescue Plan Act.
Here’s where the coming US agency and the hoped-for international effort dovetail: Their successes will hinge on data: more abundant data, more granular data, just more. In the mid-20th century, the inaccuracy of weather forecasting was the butt of late-night TV jokes. What made it a reliable endeavor was deploying data-collection devices—satellites, Doppler radar, weather balloons, automated surface-observing systems—and achieving the supercomputer processing power and graphical systems to understand and represent the results.
The data-collection devices that could help us scan the horizon for pandemics already exist. (You might be reading this on one.) Mobility data, purchase records, search terms, the words you use in tweets—all represent information that can fuel predictive tools. Public health doesn’t yet do a good job of accessing that data, collating it, and analyzing it. The channels for getting to it haven’t been carved out even in rich countries. In the Global South, the problem is worse.
“There’s so much heterogeneity in the underlying capabilities of various countries and places,” Rivers says. Obtaining that data to help a country ring alarm bells, let alone contribute to global forecasting, “might even be a matter of moving from paper reporting to digital reporting,” she adds. “It’s hard to see how you can skip to the end and have an advanced radar system without first attending to those basic pieces, when each of those pieces in each jurisdiction is a big undertaking.”
Take test results, for instance. It would be desirable to plug in the results of any diagnostic tests done during health care visits, to sort out whether a wave of respiratory infections is being caused by a common virus or a new strain. But so many people lack access to health care that diagnostic data might have limited predictive power. On the other hand, most people use sewage systems—where they exist—and wastewater sampling can detect pathogens without intruding on individual privacy or forcing the construction of interoperable record systems.