An image from a Landsat satellite of Brazil, where the Amazon flows into the Rio Negro and Solimoes River. Satellite imagery like this will be coupled with epidemiological data, meteorological data, and Internet data streams to identify conditions that could potentially lead to disease outbreaks.
By Nick Generous
Public health is like your plumbing: You don’t notice it until it’s broken. And when those safeguards and policies that are put in place to keep our communities healthy and strong are broken, the results can be devastating.
Take, for example, Haiti in 2010. Ten months after a 7.0-magnitude earthquake, a cholera outbreak spread through the country. By the end of 2011, more than 500,000 people had been infected and 7,000 were dead from the disease. In West Africa, the Ebola virus killed more than 11,000 people from 2014 to 2016.
While developed countries like the United States have better public health infrastructure to quell epidemics, we are not immune. For instance, the CDC estimates that influenza has led to between 12,000 and 56,000 deaths annually since 2010.
Better tracking of infectious diseases can help us improve disease prediction and, consequently, more quickly stop their spread. At Los Alamos National Laboratory, we’ve been using mathematics and computer modeling since the early 2000s to do exactly that. It’s easy to see how tracking diseases and stemming their spread are vital to national security. Diseases don’t care about boundaries. They don’t respect borders, and they aren’t governed by political ideology. All it takes to spread an infectious disease is for an infected person to carry it from one place to another. In today’s globally connected society, that’s all too easy.
To help prevent disease outbreaks in the United States, we need to improve public health all around the world, not just within our own borders. Better disease tracking—and, more importantly, forecasting—can help us do this. If disease outbreaks could be forecasted like the weather, communities could set up protective measures to mitigate their impact. Just as hurricane warnings spur people living in coastal areas to board up their windows, a disease outbreak forecast could alert vulnerable populations and help them take necessary precautions to keep the disease at bay.
It’s not as far out as it might seem. At Los Alamos, we’ve been using internet data sources, such as Wikipedia, Twitter, and Google, along with anonymized data from doctors’ offices and other clinical settings, to feed our disease-forecasting mathematical models. We’ve had some success (for example, we’ve been able to accurately predict the timing and severity of the last two flu seasons), but realized that we could improve our models by incorporating demographic and environmental data such as moisture, land use, standing water, temperature, and deforestation, and in turn identify additional factors contributing to mosquito-borne diseases.
That’s where satellite imagery comes in. We’re partnering with Descartes Labs, a company born from Los Alamos scientists, to use high-resolution satellite imagery to detect disease-harboring environments such as areas with high soil moisture that may be conducive for mosquito breeding. By mapping where high-moisture areas intersect with those social media signals and clinical surveillance data, we can help identify areas at risk for disease emergence and subsequently predict the potential path of the disease. This information can then inform public agencies and officials who must decide how best to allocate resources and implement mitigation strategies.
Descartes Labs collects data daily from public and commercial imagery providers, aggregating the images into a single database. Our team at Los Alamos will use the Descartes Labs Platform to correlate satellite imagery with multi-year clinical surveillance data from approximately 5,500 Brazilian municipalities for mosquito-borne diseases such as dengue, chikungunya, and Zika viruses in order to better understand how they spread. Do they hit big cities first, where congested living conditions might hasten their spread? Or do they first emerge in rural areas? Or is it simply a matter of how much standing water and moisture is present, regardless of the urban or rural setting?
We can also use satellite data to quantify vegetation health in local areas and correlate those with the amount of water and the number of mosquitoes. We can narrow it even further to look at land use: Is it a forest? Is it urban? Is it agricultural land? Does deforestation in the Amazon result in more irrigation, which results in more stagnant water, which means a proliferation of mosquitoes and the diseases they carry?
Right now, we don’t know. Descartes Labs’ satellite data platform coupled with Los Alamos’ unique capabilities in remote sensing, high performance computing, epidemiology, genomics, and mathematics will help us find the answers. Furthermore, this satellite imagery allows us to remotely look at extremely small geographical areas. Not only can we look at a city, but we can look at a specific neighborhood within that city—giving us insights that have previously been inaccessible.
To help us determine whether our mathematical models are accurate, we’ll use retrospective analysis—which uses historical data (where we know what happened)—and see if our models accurately predict the outcome. This will then give us confidence that our models could do prospective analysis: predict the future.
We’re expecting to have our initial results this fall. Whatever we learn, it will get us one step closer to better understanding how diseases spread—and, ultimately, help save more lives. That will mean a brighter and more secure future, not just for the United States, but for the world as a whole.
Nick Generous is a digital epidemiologist at Los Alamos National Laboratory. Other members of the project team include Sara Del Valle, mathematical epidemiologist; Geoffrey Fairchild, computer scientist; Nidhi Parikh, computational scientist; Carrie Manore, mathematician; Jessica Conrad, mathematical epidemiologist; and Amanda Ziemann, a remote sensing scientist.
This article originally appeared in Scientific American.