As an expert on the transmission of infectious disease, Columbia University professor Jeffrey Shaman is at the forefront of studying the coronavirus. The Mailman School of Public Health environmental health sciences professor and Data Science Institute affiliate member helps officials understand how COVID-19 spreads and what they might do to mitigate its transmission.

Shaman has been interviewed by dozens of journalists, and The New York Times shared a coronavirus database with him to help reporters evaluate the data. The paper published two major stories as a result of that collaboration. For the first, which ran March 20, Shaman used transportation data from the U.S. Census Bureau to model how the virus could spread across the country. And for the second story, which ran March 22, he examined why travel restrictions in China did not prevent the spread of the virus.

But perhaps most significantly, he is senior author of a study published by Science on what he refers to as the “stealth transmission” of the virus. For every confirmed case of the virus, there were an estimated six additional people in China with undocumented infections prior to the implementation of control measures. Though about half as infectious as documented cases, people with undetected symptoms were responsible for about 79 percent of China’s coronavirus cases.

“Undetected cases expose a greater percentage of the population to the virus,” Shaman says. “And these stealth transmissions will continue to present a major challenge to the containment of this outbreak around the world.”

The authors’ findings were based on a data-intensive computer model of the coronavirus outbreak and used mobility data and reported infection rates to simulate the spread of the virus within China. The model also estimated the contagiousness and amount of undocumented infections in China before and after the government instituted a travel shutdown and other control measures in Wuhan.

Shaman has used similar modeling approaches to forecast seasonal outbreaks of the flu and conducted field work in New York measuring undocumented respiratory infection rates. He learned that only about 20 percent of people with influenza ever see a doctor and are diagnosed. That made him suspect that the same scenario had played out with the coronavirus in China and he collaborated with co-authors from London and Beijing to gather data from China and adapt the influenza model. The researchers conducted their study from January 10-23, before the Chinese government imposed a travel ban and started aggressive testing, and from January 24-February 8.

“I don’t consider this a big data study,” Shaman says. “We used aggregated mobility data about the movement people between cities and scaled the data accordingly.”

The authors simulated transmission within and between 375 Chinese cities. They categorized infections according to documented or undocumented cases and ran an inference problem to estimate essential parameters. The essential questions were a) what percentage of the Chinese people experienced COVID-19 infections? and b) what fraction of infections were undocumented and how contagious were they? They found that people who had the virus but went undiagnosed, probably because they had mild symptoms, were the source of the majority of China’s documented COVID-19 cases in the early period of the outbreak.

Given the high probability that the virus will follow similar “stealth transmission” patterns in other countries, the study received global attention.

Shaman says protective measures, including quarantines and physical distancing, coupled with more active testing could help stem transmission of the virus, but that each country is facing a different yet unprecedented predicament.

“We haven’t seen something this catastrophic since the Spanish flu of 1918 and it’s requiring sacrifices that we haven’t seen since World War II,” Shaman says. “There are going to be enormous disruptions. There’s no easy way out.”

— Robert Florida