Innovation/ Field Note

On the Frontline in Fragile Countries: Using Data to Prioritize Covid-19 Resources and Save Lives

Photo courtesy of Esteban Sacco in Juba, South Sudan.

In Juba, South Sudan, Esteban Sacco, deputy head of office at the United Nations Office for the Coordination of Humanitarian Affairs (OCHA), spent four hours recently on a call with partners, trying to reach a consensus on where to prioritize spending $5 million in resources to fight Covid-19.

In the end, the group selected four urban areas, but deciding where the risk was highest was a matter of educated guesswork—“building the plane while we’re flying it,” as Sacco puts it.

“What we are doing is like giving the patient a generic antibiotic to fight a particular disease,” he said. “If we are able to turn to specific data, and at the most local level possible, we can do a better job prioritizing. Where should we set up isolation or treatment centers? Where should we distribute PPE? Where to do more prevention or social mobilization? Even with imperfect data, we will make better guesses.”

That in a nutshell is the urgent goal of a partnership between The Rockefeller Foundation, OCHA’s Centre for Humanitarian Data and Johns Hopkins University’s Applied Physics Laboratory (APL). Under a $500,000 year-long grant that began May 1, a deeply experienced and committed team is working hard to save lives and inform humanitarian interventions by closing data gaps in fragile states and developing models to anticipate the location, scale, severity and duration of Covid-19 outbreaks.

“There are today more than 450,000 cases in the 25 countries in which the UN has major humanitarian operations, more than a threefold increase compared to last month,” said Leonardo Milano, the Centre’s Predictive Analytics Team Lead. “The need is critical for reliable models which can support decisionmakers and highlight the most effective policy interventions.”

  • When making decisions about Covid-19, days matter. The OCHA team in partnership with Johns Hopkins are building models to give those on the frontlines of the Covid-19 response timely and tailored information they needed to make smart decisions now.
    Evan Tachovsky
    Director & Lead Data Scientist, Data & Technology, The Rockefeller Foundation

Data Translates into Better Decisions

APL’s Sheri Lewis, a global disease surveillance expert with a specialty in lower- and middle-income countries, echoes that sentiment. “We’re not just talking about numbers on a page,” she says. “Data translates into knowledge, which translates into better-informed decisions. And anything that we can do to help someone make a better-informed decision, especially at a time of crisis, is a small victory.”

OCHA’s Manu Singh, a predictive analytics consultant, shows a “decision tree” used in modeling.

The project begins with four countries—Sudan, South Sudan, Afghanistan and Democratic Republic of Congo—where Western models anticipating the speed and spread of Covid-19 infections are not very useful. OCHA is collecting the data and APL is creating the models.

“When you look at the models developed for Europe, the U.S., or China, you are essentially looking at key variables like age or medical conditions,” says Sacco, who has served in hotspots for most of his career. “We want to put other data inside our models to better understand the condition of people in targeted countries.”

In Afghanistan, for example, where the team developed the initial model, they found that Covid-19 crept in from the rural west instead of moving out from a city as it typically did in other locations. The reason? People were crossing the border from Iran where the outbreak was intensifying. Understanding where those communities were headed next would help predict the course of the virus.

And in South Sudan, the world’s newest country after gaining independence from Sudan in 2011 and demographically the youngest nation in the world, the median age is 19 years old. Teenagers are normally considered low on the risk category for Covid-19, but one question a South Sudan model wants to ask is what happens to risk when young women are chronically malnourished and spend hours daily cooking over indoor wood and charcoal ovens?

 

The Team Faces Multi-Faceted Challenges

The predictive project is ambitious, but the team is humble. Jason Lee, a biostatistician with a public health background who leads the partnership for APL, cites British statistician George Box—“All models are wrong, but some are useful”—in noting that the work they are creating will almost inevitably fall short of the complexities of reality.

 
 

In addition to the usual challenges inherent in model creation, this team faces others:

  1.  They are trying to collect data from places where data has often not been typically gathered and sometimes not even valued.
  2.  They are taking this on in the midst of not only the pandemic but other enduring crises, ranging from political instability and conflict to chronic poverty and food insecurity.
  3.  And they are working with governments and citizens wearied by a long menu of problems. “Communities here have learned to respond to threats only if they are imminent,” Sacco says. “If they hear a conflict is coming, they don’t care. When they actually hear the shooting, then they run.”

Still, the potential rewards are significant when balanced with limited resources. Covid-19 has overwhelmed some of the world’s most robust healthcare systems, which highlights the stresses put on more vulnerable countries. “There are maybe 100 hospital beds in all of South Sudan,” Sacco notes.

The model can simulate the effect of non-pharmaceutical interventions and inexpensive local solutions, such as closing particular roads or schools, or mandatory mask wearing.  “Even limited data will help with that kind of decision-making,” Lee says. “We want to make marginal gains that add up over time. And if people begin to understand the value of collecting this data on a routine basis, that will be a great outcome also.”

And though the data they seek to include in the model is on the most local and granular level possible, the benefits defy borders. “It is a truism that a public health emergency anywhere is a public health emergency everywhere,” says Lewis. “Diseases can and will spread. So it is critically important that everybody have the capacity to collect data and perform disease surveillance.”

  • It is a truism that a public health emergency anywhere is a public health emergency everywhere. Diseases can and will spread. So it is critically important that everybody has the capacity to collect data and perform disease surveillance.
    Sheri Lewis
    Program Area Manager, JHU APL

“Our Lives Are at Stake”

The OCHA effort is aimed not only at creating predictive models, but at providing a peer review process to evaluate the models for their technical readiness and consider any ethical concerns. They have established a framework that can be used to assess the models.

“The Covid-19 pandemic has brought into stark focus the need for data and the value of models to inform response strategies,” says Sarah Telford, the Lead for OCHA’s Centre for Humanitarian Data. “We are all learning more about epidemiology than we ever imagined, from the reproduction rate to flattening the curve. And for good reason: our lives are at stake.”

South Sudanese refugees practice social distancing as they wait to access a food distribution at Kakuma camp. Photo credit: UNHCR

The Centre began its work in 2017 with the broad goal of increasing the use and impact of data in the humanitarian sector. It manages the Humanitarian Data Exchange platform, which currently includes almost 20,000 data sets from over 250 partner organizations covering all active humanitarian crises.

Covid-19 is not the first health crisis that the Centre has worked on. In response to the 2018 Ebola outbreak in the Democratic Republic of the Congo, the Centre helped its partners analyze data about rumors, looking for insights that could help responders meet community concerns.

In some ways, systems being put in place during the Covid-19 pandemic will be even more beneficial in the next emergency, notes Jessica Dymond, a senior research scientist at APL with expertise in synthetic biology, emerging biotechnology and biosecurity, and a lead on the project. “What a lost opportunity it would be not to take lessons from this,” she says. “I hope we are demonstrating how important data is in a crisis.”

Dymond says she often thinks about how her two young children will look back on this pandemic. “What I’d like, in 20 or 30 years’ time,” she says, “is to be able to point out to them data and modeling capabilities that we have in place, and that wouldn’t have been there without the work we are doing now.”

The Centre for Humanitarian Data

OCHA coordinates the global emergency response to save lives and protect people in humanitarian crises. Advocating effective and principled humanitarian action by all, for all.

Learn More

Related Updates

Jun 09 2020
Blog Post
Anticipating Humanitarian Crises in the Age of Covid-19
More
Dec 30 2014
Blog Post
Big Data Offers New Opportunities for Community Resilience
More
Jul 22 2015
Blog Post
Developing Guidelines for Humanitarian UAV Missions
More
Back to Top