Satellite remote sensing data enhances projections of malaria outbreaks

A team of researchers at South Dakota State University is currently collaborating with public health organizations in Ethiopia to predict and prevent malaria outbreaks.

The research team uses earth observation data sets derived from satellite remote sensing, said Mike Wimberly, SDSU natural resource management professor and principal investigator.

“The advantage of [satellite remote sensing] is that satellites can take these environmental measurements everywhere in the world,” Wimberly said. “So, in a place like Ethiopia that doesn’t have a lot of weather stations, for example…we can really leverage these satellite measurements to study anywhere in the world.”

Wimberly said the goal is to help make emergency responses to malaria more proactive and efficient. This could ultimately lead to malaria elimination over the next couple decades.

However, to conduct the research, epidemiological data is also needed. This is actually tracked somewhat closely in Ethiopia, Wimberly said. This data comes from the collaborators in Ethiopia.

The research is a collaborative effort between faculty at SDSU and collaborators in Ethiopia including the Health Development and Anti-Malaria Association (HDAMA), a non-governmental organization, and the Amhara Regional Health Bureau, the government agency ultimately responsible for mobilizing emergency response to malaria and other epidemics. The National Institutes of Health (NIH) have funded the research done by the Geospatial Sciences Center of Excellence at SDSU.

Wimberly and his research team have been working with Yi Liu, an SDSU associate professor specializing in software architecture and development, since 2009 to develop the software used for the research.

Liu said since 2009 she has had several undergraduate and graduate students assist in the software development.

The process of developing the software started with the “user requirements,” or what the researchers wanted to be able to do with the software. 

Several meetings were held to prioritize needs and develop prototypes based on those needs. Then came designing, developing, coding and testing the software.

“Testing is a very important part because, of course, we would like to deliver something to our users without major bugs and we do not want to crash their system,” Liu said.

The software Liu and her team developed is flexible enough to support future changes so that when they modify it, it will not hurt the entire system, Liu said. 

The team designed two separate software for the users in Ethiopia and the users at SDSU.

“We are currently integrating two databases together so that we can do the malaria prediction from the prevention data plus environmental data,” Liu said.

Through the software, the research team works on epidemic detection and early warning. They model an expected number of malaria cases, track those cases and see if the cases are exceeding what was expected. They also project future outbreaks multiple weeks in advance based on delayed responses to environmental variables, Wimberly said.

The highlands of Ethiopia and parts of east Africa are considered epidemic prone, Wimberly said.

Typically, malaria does not occur in the dry season because mosquitoes need water to breed. The main malaria season is at the end of the rainy season, during North America’s fall, because it creates an environment that is wet but warm with sunny conditions, which is prime for malaria transmission, Wimberly said. 

The research process has not only required data collection, time studying the data and understanding the connections between the environment and disease, but it has also required building up social connections and partnerships with non-governmental organizations as well as public health institutions in Ethiopia, Wimberly said. 

“The challenge is that, to implement it, this sort of project requires building,” Wimberly said. “It really requires partnerships between researchers and public health practitioners which are not always easy to keep going because researchers and public health practitioners often have different agendas.”

Chris Merkord, a postdoctoral research fellow working for Wimberly, agreed that building partnerships in Ethiopia is imperative.

“I think that’s the missing link that a lot of projects don’t have–they don’t have that strong connection to people who are charged with making decisions on the ground,” Merkord said. “That’s all Mike. He’s seen that need from the beginning and made sure we have that foundation to build upon.”

Wimberly estimates he has been to Ethiopia roughly 10 times over the past eight years. The research team visited Ethiopia in February and their Ethiopian collaborators will be visiting SDSU in the summer.

“To really get things done you need to show up,” Wimberly said. “I think another part of it is just staying in the game.”

Most of the time during the team’s visits to Ethiopia is devoted to meetings and getting feedback from their collaborators in Ethiopia.

“We’re scientists, and so we’re thinking about things as scientists do,” Merkord said. “They’re thinking about it in terms of how to use it to help them and their job to ease disease burden on our society and that’s a big responsibility.”

The process has been long and slow, Wimberly said. However, progress has been made.

“We’ve gotten to the point now where we’re actually making forecasts and pushing them out to our public health partners for a set of pilot districts in the Amhara regions which are in the Ethiopian highlands,” Wimberly said.

The forecasting is still not perfect, Wimberly said. It is a continuous development process. They go through stages of developing a model, making predictions with it, developing products, reports and visualizations based on those predictions, getting feedback and updating the models.

“It’s like our airplane isn’t completely built yet, but we’re still trying to fly with it,” Wimberly said.

Predictions, however, are never perfect. It’s kind of like forecasting the weather. There’s the potential to get it right a lot of the time, but it’s never going to be perfectly right. So, the goal is to make something that’s useful, Wimberly said.

Though there weren’t many outbreaks during the first season of testing last fall, the results were encouraging, Wimberly said.

“We did get early warning indicators of two major outbreaks in the region so that was a promising sign,” Wimberly said.

Wimberly also works with Geoffrey Henebry, co-director of the Geospatial Sciences Center of Excellence.

Wimberly, Henebry and Merkord agree that culture has been one of the most notable challenges.

“You’ve got distance and language and culture, which is extremely different,” Wimberly said.

A lot of the challenges are not glamorous, Merkord said. There’s a lot of small software challenges along the way. Long distance communication has also been a challenge for the team. 

“There’s the straightforward challenge of anytime, when you’re communicating by email with someone in English there’s lots of vagueness, uncertainty, ambiguity,” Henebry said. “When you’re communicating with someone where English is the second language that’s that much more complicated.”

Internet access has increased in Ethiopia over the past few years, Henebry said. This increased communication has made the goals of the research more realistic.

“When we first started there in 2009, it was very difficult to get online,” Henebry said. “And that’s been an important consideration in terms of what is feasible because when we first got there malaria information was communicated at all levels by the telephone.”

Though there are challenges, Ethiopia is a great country to work with, Merkord said.

“I can’t generalize to the entire continent, but most of them are really great people to know—kind and generous and hard working, very intelligent people,” Merkord said. “It’s a great place to work and they’re great people to work with.”