Malaria and mobility

Map of Kenya showing malaria recep­tivity of var­ious settlements.

According to the World Health Orga­ni­za­tion, more than 600,000 people died from malaria in 2010, with a majority of those deaths occur­ring among African chil­dren. The infec­tion is caused by the par­a­site Plas­modium fal­ci­parum and is trans­mitted via mos­quito bites. As you might imagine, where there are more mosquitos—and fewer dol­lars for public health programs—there is more malaria.

But thanks to increased pre­ven­tion and con­trol mea­sures, malaria mor­tality rates fell between 2000 and 2010 by more than 25% glob­ally and by 33% in Africa, according to the WHO.

Before joining the Col­lege of Com­puter and Infor­ma­tion Sci­ence, research assis­tant pro­fessor Nathan Eagle lived in Kenya for two years working as a Ful­bright Pro­fessor at the Uni­ver­sity of Nairobi. Along with his wife Car­o­line Buckee, a com­pu­ta­tional epi­demi­ol­o­gist at the Har­vard School of Public Health, Eagle saw first-​​hand how malaria can effect the lives of people living in high-​​risk communities.

Even if you erad­i­cate malaria in a par­tic­ular region of Kenya,” said Eagle, “the ques­tion is, will that region be able to stay erad­i­cated?” The answer, he said, lies partly in human mobility data. After all, no one stays in one place any­more and trav­eling between regions of varying malaria rates could under­mine the efforts of a single region.

In an article to be pub­lished in the journal Sci­ence on Friday, Eagle, Buckee and a team of epi­demi­ol­o­gists and data sci­en­tists report an ana­lyt­ical method that could inform future public health pro­grams in devel­oping more effec­tive erad­i­ca­tion strategies.

Every day, mil­lions of Kenyan mobile phone subscribers—just like everyone else in the world—leave traces of their where­abouts when local cell towers transmit their calls. Eagle and his team decided to “take this data and use it in a way that improves the lives of the people who are gen­er­ating it.”

If a subscriber’s phone con­nects to a cell tower in Nairobi one day and then to a cell tower near Lake Vic­toria the next, the researchers infer that this indi­vidual trav­eled between the two places over the last twenty-​​four hours (it being rather unlikely that the phone would make that journey on its own). The team looked at these mobility traces for nearly 15 mil­lion sub­scribers and used them to gen­erate a global snap­shot of travel among the entire population.

By cou­pling this infor­ma­tion with epi­demi­o­log­ical “preva­lence data,” stake­holders can get a more com­pre­hen­sive pic­ture of malaria trans­mis­sion across the country.

We first qual­i­fied the aggre­gate pop­u­la­tion behavior,” said Eagle, “looking at the nation as a whole.” They then zoned in on par­tic­ular regions, sep­a­rating them based on envi­ron­mental char­ac­ter­is­tics that deter­mine the like­li­hood of dis­ease trans­mis­sion. Two fac­tors will inform whether a subscriber’s travel is likely to move the par­a­site, in addi­tion to the person, from one place to another: the char­ac­ter­is­tics of their home and those of the place they are visiting.

How do you make sure that if you do get rid of malaria in one par­tic­ular vil­lage that you’ll have a high prob­a­bility that it stays erad­i­cated?” asked Eagle. This work pro­vides a tool for policy makers to decide whether com­plete erad­i­ca­tion efforts in a single region make sense economically.

While they can be very con­fi­dent that they can erad­i­cate it in the short term, if they don’t take [mobility] into account,” he explained, “they may be wasting their money.”

While some regions of Kenya are so iso­lated from both malaria and travel that it is unlikely they’ll be rein­fected after com­plete erad­i­ca­tion, most of the nation might be better served through national erad­i­ca­tion approach rather than a local one, said Eagle.