3 ways advances in disease modeling will help end malaria
Huge progress has been made in the last 15 years in reducing the number of malaria cases and deaths, and many countries have now declared elimination goals. As the number of cases go down, malaria no longer is distributed evenly nation-wide, but becomes limited to pockets in communities that are often very tricky to reach operationally or in populations with limited access to health care.
I lead our team’s work on applied epidemiology, which means developing modeling and analytics technologies focusing on malaria elimination that are used to guide decision-making and operations in our priority geographies. This includes work with the malaria modeling consortium on ways to stratify malaria in a place and to determine the optimal package of interventions, genotyping parasites to track drug resistance and tell whether infections are from a certain area versus imported from somewhere else, using cell phone data records to assess population mobility, and making sure all these exciting new approaches do not stay as purely academic exercises – rather that they combine together in an accessible way to help inform decisions by national malaria control programs.
The global polio eradication program pioneered ways to drive advances in GIS mapping technologies to make sure vaccination teams in northern Nigeria were able to reach every last child with a vaccine. The tools used for polio are directly applicable to malaria elimination programs on the ground as they aim to reach every last parasite.
The world has a chance to learn from previous epidemics and understand the importance of disease surveillance systems to identify and respond to a broad variety of infectious diseases. In these settings, technological innovation is even more important to help guide where to focus scarce resources.
The malaria team at the Bill & Melinda Gates Foundation considers surveillance the backbone of any malaria elimination program, and modeling and analytics techniques help take data for decision-making even further. These techniques take information on where disease exists in a place and then help us classify areas by the vector behavior and mobility of the population, test assumptions about what is the appropriate combination of interventions, and tell us whether or not those interventions are working. This is an important step on the path to eradication because it ultimately saves money for programs as they can target resources on the areas that need it most to drive down transmission, a strategy that helped lead to smallpox eradication and is currently used by the global polio eradication program.
An example is the DiSARM platform developed by the University of California San Francisco which we support in partnership with Google. There are multiple data sources needed for a malaria program to make decisions – such as case data, intervention coverage, and health facility stock information – that often exist in different databases and formats within a country. DiSARM solves the problem by rapidly pulling together and visualizing this data via user-friendly platforms. The platform then layers this data on top of risk maps through Google Earth Engine, a powerful back-end analytics engine to convert raw data into actionable information for on the ground programs. DiSARM is currently undergoing final development and is already being used by malaria programs in four countries in southern Africa to help plan, execute and monitor indoor residual spraying campaigns.
Sign up for The Optimist newsletter
2. Technological innovations on mapping
As a country sees fewer cases of malaria, the government needs to shift their focus on targeting their programs to smaller and smaller geographical scales. How do programs plan campaigns to make sure they reach every household in an area with mosquito control and the right diagnosis and treatment? By mapping the location of every household. Doing this the traditional way, making a map by physically visiting every house in an area, takes an extremely long time and is very expensive.
But advances in the technology industry are making this process quicker and easier. If you stand on the streets of Seattle and open up Google Maps you can see detailed outlines of buildings and information on the businesses that operate there. It is a very different picture if you stand on a street in rural Laos where many communities are not currently on maps. A year ago, it was not technologically possible to detect buildings in many parts of the world, but now advances in machine learning algorithms have made this possible. We are now applying this technology for malaria elimination.
A great example is the partnership we set up between DigitalGlobe, the Humanitarian Open Street Map team, and the Clinton Health Access Initiative that builds on the work of the polio global eradication program. This partnership takes satellite imagery, advances in machine learning, and the power of crowdsourcing to map exact building locations in the Gates Foundation’s priority areas for malaria elimination Southern Africa, Southeast Asia and Central America, and today they are using this information to support malaria elimination programs on the ground like planning for vector control campaigns and the optimal placement of community health workers.
3. Mathematical modeling
How do we know that interventions to target to which places? Malaria persists based on complex interactions between the parasites that cause the disease, the mosquitoes that transmit parasites to humans, the environments humans live in that make them either more or less susceptible to exposure to mosquito bites, how well different interventions work in different settings, and the governments and health systems that exist in different countries.
Mathematical models can help us make sense of all of those dynamics and help determine the optimal package of interventions for deployment. Our modeling and analytics partners, including the Malaria Modeling Consortium (MMC), are now more than ever before working together with national malaria control programs including Zambia, Haiti, Mozambique to pull together the surveillance data, information on genetic sequencing of parasites, geospatial risk maps, and location of populations so that the national programs can make data-informed decisions on how to best use their available resources to target interventions and eliminate malaria.
The partnership between the MMC, MACEPA, and the Zambia National Malaria Elimination Centre over the years has helped the program understand the settings where mass drug administration is likely to have the most impact, the cost effectiveness of different options for their National Strategic Plan, and how to stratify the country into different levels of transmission to target different interventions as cases decrease. We hope to see more of this type of partnership in the future, including continued support from the Global Fund for modeling to help support resource allocation.
Our next step is ensuring all these amazing advances in technology and analytics are harnessed and used in the countries that need them. Data-informed decision making for malaria elimination includes bringing together surveillance data on cases and vectors and health systems, risk maps, genetic relatedness of parasites, population location and their mobility patterns, and input from transmission modeling. I am currently writing this from Port-au-Prince, Haiti, where I am working with the Program Nacional de Control de la Malaria and the Malaria Zero consortium to pull together surveillance data, risk mapping, community surveys on incidence and serology, parasite sequence data and modeling to decide what combination of interventions to target to which places to eliminate malaria from Haiti. The more that Haiti, and other countries no matter if they are high or low burden, can use these approaches the better chances they will have to accelerate the timeline to elimination.