Using satellite data to predict cholera outbreaks

Thursday, January 25, 2018

I came across a nifty piece in Scientific American this week about a group of scientists whose algorithm to predict inland cholera outbreaks was unexpectedly proven to be accurate:
Quickly collecting ground data about [cholera epidemics] can be challenging, especially in chaotic locations. Yemen is a textbook case. “Yemen has massive civil unrest, people are moving around, [there is] political instability—there’s no way for us to get a single data point,” Jutla says. But satellites gave his team a way to assess the disease risk from the sky, and without being in the country.

At the American Geophysical Union annual meeting in December, Jutla presented the group’s prediction model of cholera for Yemen. The team used a handful of satellites to monitor temperatures, water storage, precipitation and land around the country. By processing that information in algorithms they developed, the team predicted areas most at risk for an outbreak over the upcoming month.

Weeks later an epidemic occurred that closely resembled what the model had predicted. “It was something we did not expect,” Jutla says, because they had built the algorithms—and calibrated and validated them—on data from the Bengal Delta in southern Asia as well as parts of Africa.
It's important to keep in mind that just because the model got this instance correct does not mean there is not room for improvement. However, it seems to have support from other experts:
“One of the things I like,” says Michael Wimberly, an ecologist at South Dakota State University, is that they are not looking “only at correlation to rainfall.” ...He says the cholera model is well grounded in hydrology and epidemiology. “They have an understanding of different types of epidemics that occur in different seasons; it’s very sophisticated."

No comments :

Post a Comment