Detect Epidemics Before They Start

Photo: Bruce Gilden

Back in May 1993, as a medical resident at the University of Arizona, Mark Smolinski volunteered for a shift with the state's Department of Health. Right after he started, Arizona and neighboring states were struck by a deadly outbreak of an unidentified respiratory illness. The young doctor found himself face-to-face with an emerging epidemic, part of a team that spent sleepless months struggling to contain the outbreak. "I was going from hospital to hospital trying to determine the patients' exposures," he recalls of his harrowing first assignment. "Almost all the cases were under the age of 30, and it had a very high mortality rate."

The researchers finally identified the culprit — which eventually infected 53 people, 60 percent of whom died — as a new strain of hantavirus. They pinned the outbreak on a confluence of ecological and social factors: Wet weather during an El Niño year spawned heavier-than-normal vegetation. That in turn fueled an unusually large population of deer mice, which harbor the virus. The victims were exposed when they rummaged through closets or gardened, inhaling dust laced with mouse droppings, urine, or saliva. The disease soon receded, but Smolinski was hooked on the rush he got from investigating outbreaks. "It seemed like a career that would never be dull," he says. "That has certainly proven true."

These days, Smolinski's business card at Google.org, the philanthropic arm of the Mountain View behemoth, identifies him simply as "threat detective." He's director of the organization's Predict and Prevent Initiative, a global health program. The 46-year-old's job is to channel money — one insider estimates up to $150 million — into projects and technologies that will help catch outbreaks like hantavirus wherever they crop up. What's even more ambitious is Smolinski's desire to push disease surveillance "two steps to the left of the epidemic curve." The strategy: Draw on Google's search acumen to predict hot spots before the first case of some imminent calamity hits the hospital.

Smolinski faces a daunting landscape. More than 300 new diseases have emerged since 1940, many the result of jumps from wild animal to human. Outbreaks are expected to increase as environmental degradation thrusts humans into ever-closer contact with wildlife and as climate change alters the life cycles of disease vectors — like El Niño did to the deer mouse. Meanwhile, older diseases are rapidly crisscrossing the planet. "West Nile, which has been around in Africa since antiquity, appeared in 1999 in New York, and in three years it spread throughout the country," Smolinski says. "Now it is one of our endemic diseases."

Yet even though the next West Nile or HIV is just a plane flight away, the global public health system remains focused on responding to diseases after they've spread. That needs to change, Smolinski says. And he's hardly alone in his thinking. "The government and organizations that care about the health of the country and world need to think very carefully about whether our current approaches to controlling disease are sufficient," says Nathan Wolfe, a UCLA epidemiologist who specializes in identifying and studying new diseases. "If they do, they'll find that we are like the cardiologists of the 1950s, waiting for heart attacks to occur." A major study in Nature this year found that, despite the SARS and H5N1 avian flu scares, the areas most likely to harbor some new outbreak — like Malaysia, hit by the Nipah virus, or Kenya, which recently had a Rift Valley Fever outbreak — were least likely to have the kind of disease surveillance systems that let health pros monitor clusters of symptoms or animal die-offs.

Number of emerging infectious diseases reported worldwide.

To change that, Smolinski plans to lend Google's in-house technology expertise to companies and nongovernmental organizations already at work in the developing world. One effort involves extending traditional surveillance closer to the source of epidemics, beginning in outbreak-prone regions of Southeast Asia and sub-Saharan Africa. With the right network in place, Smolinski says, a farmer in Vietnam might send an SMS when chickens start dying or a kid has an unexplained fever. That information alone, of course, means little — "chickens die all the time from Newcastle disease," Smolinski says. "It doesn't have to be avian flu." That's where the search company comes in. Google.org partnered with HealthMap, a Web crawler that Smolinski says can reveal outbreak patterns hidden in vast amounts of public data. (The Global Public Health Intelligence Network, a Canadian project, used similar Web-scanning software to help detect the first signs of the SARS outbreak.)

Google.org is also looking at for-profit companies inventing and manufacturing low-cost medical diagnostic tools. "Molecular sequencing methods have gotten less expensive in the developed world," Smolinski says — cheap enough to be used to identify viruses in the field. Combine that genetic knowledge with open source information — farmers' SMSes, hospital reports, climate data, and satellite imagery — and Smolinski thinks researchers can begin not only to find early outbreak patterns but also understand the conditions most likely to create the next interspecies leap.

Despite Predict and Prevent's numerous resources — several multimillion-dollar, multiyear grants, plus 1 percent of Google equity — turning emerging disease research into a predictive science in some of the poorest health care systems in the world remains a long shot at best. But the man who once served as an epidemic intelligence officer at the Centers for Disease Control and was director of a landmark National Academy of Science study on future microbial threats has a record of achieving the unthinkable in public health. As vice president of biological programs for the nonprofit Nuclear Threat Initiative, Smolinski led the formation of a disease surveillance program in the Middle East, somehow coaxing Egypt, Israel, Jordan, and the Palestinian Authority into sharing outbreak data and lab results. This collaboration has already succeeded in containing cross-border flare-ups of mumps and other infections. "One thing that became clear is that health was a great tool to get people to work together, because they are in the same boat when it comes to infectious diseases," Smolinski says. "And if you can do it there, you can do it anywhere."