How We Map Epidemics

As coronavirus cases multiply, interactive maps are helping us understand the spread of the disease—and panic about it.

One of the best examples is the . At the time, plague was wreaking havoc throughout Europe, and Fillippo Arrieta, an Italian royal auditor, spatially visualized the strategy for containing the spread of the disease in the region of Bari, Italy. On Arrieta’s map, Bari is separated from the rest of the country by a dashed line that represents a “cordon sanitaire.” Within the cordoned-off province are two smaller areas, separated from the rest by a thicker line. The observant reader can see large “D” letters in the top right area, meaning that the province is infected by the plague.

However, according to Koch’s research, the first truly detailed spatial study of an epidemic did not appear before 1797 (approximately) and the publication in the Medical Repository of Valentine Seaman’s maps of the yellow fever outbreak in New York City. Seaman overlaid the location of yellow fever cases— dots on the map below—with the position of  dumping areas and sewage sites in lower Manhattan. He marked these sites with a thick letter “S.” Reflecting upon his observations, Seaman concluded that the deadly outbreak was linked to those areas and their putrid emanations.

Valentine Seaman, An Inquiry Into the Cause of the Prevalence of the Yellow Fever in New York, in the Medical Repository, 1797. (Brian Altonen)

Even if Seaman’s theory was not quite right—yellow fever is carried by the mosquitoes that were breeding in these waste sites—disease mapping was born. Over time, technology improved and disease-related data became more available, according to Koch. However, it was only when an epidemic of cholera massively hit Europe, and especially the U.K., in the mid-19th century that disease maps exploded.

“And then it was… a watershed,” Koch said by email.

John Snow, Plan Showing the Ascertained Deaths from Cholera. The black bars represent deaths from the disease. (Wellcome Collection online archives)

John Snow’s famous map of the London cholera outbreak of 1854 remains the most popular example of this boom in spatial visualization of the cholera epidemic, as Britons desperately tried to understand its roots. Snow overlaid the location of the casualties with the position of water pumps in the city, and from there correctly deducted the water-borne origin of the disease.

He was, however, far from being the only one mapping the outbreak.

Richard Grainger, Cholera Map of the Metroplis. 1849, 1850. Via the Wellcome Collection online archives.

Sections showing the relative intensity of the attack of cholera at the various levels along the lines marked on the cholera map. (Wellcome Collection online archives)

Among the trove of mid-19th century London cholera maps is Richard Grainger’s, who hypothesized a link between the disease and altitude. Grainger mapped the city very precisely, drawing all its districts and sub-districts, and added to his map the location of sewers and wells. He overlaid elevation information and shaded areas according to the intensity of the cholera outbreak: The darker the shade of blue, the more devastating. And, in fact, Koch writes in his book, the masses near the banks of the river Thames in South London struggled with overall less clean air than Londoners who lived higher up.

Maps in the time of big data

Fast-forward a couple of centuries: The invention of the computer provided an electronic support to generate maps faster. Jump ahead a few additional decades, and the internet enables data-gathering and transmission at an ever-increasing velocity. Computers become more efficient and their data processing capabilities exploded, resulting in the creation of geospatial models that enable health care officials to understand where an ongoing epidemic might hit next, and identify those most at risk. These models make public health interventions possible.

Geraghty said when she joined Esri, the public health community was very familiar with GIS. “They had an understanding and they had been mapping, but they had been doing it with desktop tools, not web-based GIS,” she said.

As a result, larger datasets are now more easily available, and GIS users can build their own prediction models based on them. Among the trove of information readily accessible is census data, plane or ship routes, and even social media content.

In 2016, the U.S. Centers for Disease Control and Prevention used Esri’s products and expertise to monitor the diffusion of the Zika virus. Zika is spread by the Aedes mosquito, and the survival and reproduction rate of the insect is tightly linked to five variables, according to Geraghty: temperature, precipitation, land use, population and elevation. After analysis, researchers were able to identify areas of the world where the mosquito could easily live, and they cross-referenced these results with census data. Zika is particularly dangerous to pregnant women, and the census data overlay enabled the researchers to identify zones with the largest at-risk populations. This enabled effective policy-making and testing, as locals were encouraged to use insecticide and larvicide, among other means, to limit the spread of the disease.

This Esri dashboard is no longer available. Picture as seen in Healthcare Magazine, in 2016.

A dashboard (as seen in the picture above) monitoring the number of Zika cases in the United States was made available to the public. The darker the shade of red, the more cases identified in a given state.

On its end, the company Metabiota has accumulated and cleaned data on 2,400 outbreaks since its creation in 2008, according to Nita Madhav, the company’s chief executive officer. Its epidemic tracker is available to the public, but most of the firm’s modeling capacities are only available to its clients, which include the CDC and the U.S. Agency for International Development.

“This [data on previous epidemics] can actually help inform future decision-making,” Madhav said, “and can show us that these epidemics should not come as a surprise. This is something that happens frequently over time.”

The company assesses when and where outbreaks are more likely to occur, and, Madhav said, is currently “nearcasting”—projecting the short-term evolution—of the coronavirus. Metabiota also came up with a way to measure “fear,” a metric it calls a pathogen “sentiment score.” To calculate that score, Madhav explained that researchers combine existing data regarding an outbreak—like mortality rate, for example—and put it into a “scoring algorithm.” The results available to the public are ranked on a rather broad scale: the novel coronavirus “sentiment score” is defined as “high.” More granular results are available, but they are not released on the company’s website.

The John Hopkins dashboard is itself built on top of multiple data feeds. It uses CDC, WHO, ECDC and DXY reports – Chinese, American and global data sources. The resulting map is only as good as the data sources get, however, and tracking an epidemic can be tricky: Cases sometimes go unreported.

Geraghty, at Esri, regularly stops and reflects on where the field might go next.

“GIS is moving in a direction that is maybe more democratized.” she said. “There is always a need for these GIS professionals, but a lot of people whose job is totally different and need maps to make decisions don’t have to have the expertise. They just need to know enough about mapping.”

Powered by WPeMatico