Why you should care
Figuring out the key gaps of unreliable public health data could get us one step closer to effectively tracking diseases in poor countries.
Everywhere death is a mystery, but in poor countries it’s especially unknown. Record keeping is poor, and there are relatively few health care professionals to document incidence of disease. People are sick, but from what?
“To make change, you have to know what you’re trying to change,” says Abraham Flaxman, a 36-year-old who is trying to instill some order into the house of death and disease. His contribution? A computer program that cleans up the often sloppy data on hepatitis C, cholera and other diseases that plague the developing world, allowing researchers to monitor the spread of these ailments and stop them in their tracks. “If you look at the way health differs throughout the world, you see great disparities, and they don’t have to be there,” says Flaxman, an assistant professor of global health at the University of Washington’s Institute for Health Metrics and Evaluation.
It’s been used by everyone, from health workers to the mayor of Cali, Colombia, who used the program to see when violent crimes occur.
The software, which is funded by federal grants and other sources, is called DisMod-MR. It works by analyzing information such as a country’s gross domestic product and incidences of certain health issues along with other data, then estimating a disease’s progression, even in regions with conflicting or no reports. It’s been used by everyone, from health workers — allowing them to jump in well before the word “epidemic” ever gets uttered — to the mayor of Cali, Colombia, who used the program to figure out when violent crimes occur. All told, hundreds of researchers around the world use Flaxman’s program to track diseases such as malaria, tuberculosis and obesity — and their findings have been published in prestigious journals such as The Lancet. And Flaxman aims to use DisMod-MR to help prepare for some of Ebola’s saddest consequences — its worsened effects on preventable diseases like diarrhea and malaria.
Named one of MIT Technology Review’s 35 Innovators Under 35 in 2012, Flaxman is known as “the Nate Silver of public health,” David Rein, a principal research scientist in the Public Health Research Division of NORC at the University of Chicago, tells OZY. And in that sense, he’s part of a big data movement that is growing globally and is being applied to everything from public health to housing. Just as Silver applies models to different polls to forecast elections and baseball games, Flaxman applies DisMod-MR to studies of different diseases to compare their effects — an important contribution, Rein says, that helps policymakers design interventions and allocate resources to those who most need them.
Small and slightly built, Flaxman seems to have the kind of meticulous personality the task requires. His kitchen is spotless, and when he reads, he barely budges except to flip a page. He rocks a grungy, geek-chic look with slightly disheveled, shoulder-length hair, wire-framed specs and a penchant for checkered shirts. (It’s fitting that he lives in Seattle.)
But unlike the prototypically awkward, shuffling coder, Flaxman speaks in confident earnest.
But unlike the prototypically awkward, shuffling coder, Flaxman speaks in confident earnest. He credits his Reformed Jewish upbringing with planting the seeds of social consciousness early on. He caught the coding bug before he even set foot in kindergarten, at a computer camp where he wrote programs that drew flowers onscreen. Then he kindled his passion for theoretical mathematics in high school. While Flaxman went on to major in mathematics at MIT and earn a Ph.D. in the subject at Carnegie Mellon University, he, like many sullen Gen Xers, also wanted to withdraw from the world. (Naturally, theoretical mathematics offered the perfect retreat.) He also flirted with computer science and landed a postdoctoral fellowship at the Microsoft Theory Group, which tackles fundamental problems in math and computer science — without worrying about whether its research will contribute to the company’s bottom line.
But ultimately, the Iraq War in 2003 beckoned Flaxman from the ivory tower. “I felt like the world needed people to be engaged,” he says. This time, he wanted to continue toiling away in data science but to apply it “where it would matter.”
As fate would have it, the Institute for Health Metrics opened its doors right as his postdoc at Microsoft drew to a close. Flaxman’s first project: to revamp an older, clunkier disease-tracking computer program called DisMod II to draw conclusions from scientific journal articles, local health department records and Global Burden of Disease data, which comprises World Health Organization and United Nations estimates for populations, births, deaths and causes of death, as well as WHO estimates for specific diseases around the world. But as Flaxman combed through the data, he was “shocked.” Studies that looked at the same disease in the same populations had generated wildly different results. Other regions completely lacked data.
Flaxman rebuilt DisMod II from the ground up to create DisMod-MR, at a cost of hundreds of thousands of dollars. After using it to generate preliminary findings, he realized the program would be more useful if it could make predictions for individual countries, not just geographic regions. “That meant redoing everything,” says Flaxman, who crammed about two weeks’ worth of work into a single weekend. Of course, he hardly slept, and he alternated between coding and leafing through the book A Game of Thrones. It was a huge step forward, though now he wants to calibrate DisMod-MR to make predictions at provincial, state and county levels, too.
But his data grabbing isn’t stopping there. Since Flaxman knows public health data relies heavily on causes of death, he’s also developing a computer program to tackle the problem of inaccurate, incomplete death certificates in the developing world. Typically, a doctor reviews information gathered from a deceased person’s relatives and friends — a process known as a verbal autopsy. Flaxman’s program analyzes data about a wide variety of people with known causes of death to calculate correlations between symptoms and causes, using them to infer how someone died. After applying the program to 2,000 randomly selected Iraqi households, Flaxman and a colleague at UW reported in PLOS Medicine that most deaths could be traced to violence, though about a third resulted from failures in health, sanitation and other infrastructure systems.
Yet Flaxman wants to do more than just revel in data. He wants to develop tools that will make it easier to broaden access to health care in the poorest countries, which have little, if any, disease data. “His translation of [higher mathematics] to public health is very innovative in our field,” Rein says. “It allows you to make predictions.”
Photography by Jordan Stead for OZY