Intellectual Ventures Lab

Modeling the Eradication of Malaria

Despite decades of attempts to control malaria, the disease still afflicts some 250 million people every year and claims the lives of about one million, mostly children. The parasite that causes malaria has shown stubborn resilience against the most power­ful antimalarial drugs, and the mosquitoes that transmit the parasite have similarly grown resistant to insecticides. Although there is great hope for an effective vaccine, none is yet available.

At Intellectual Ventures, we believe history shows that trying to control malaria is an insufficiently ambitious goal. We in the scientific and technical community should instead develop tech­nologies and strategies that can be used to completely eradicate the disease. Much of the progress we make toward eliminating malaria will also be directly useful in exterminating other infec­tious plagues of humanity, such as polio and tuberculosis.

Toward this goal, a team led by Dr. Philip Eckhoff is de­veloping a completely original computer model that calculates not only how malaria spreads in a particular part of the world, but also how it will respond to a deliberate suppression cam­paign. The goal of this model, more ambitious than any similar software ever attempted before, is not just to understand and control the disease, but to stamp it out completely.

The model incorporates an enormous range of input data. The software contains biological information on the behavior and reproductive rates of the parasites and the mosquitoes that carry them, as well as information on infection patterns and immune responses among humans. Data from a wide range of sources on where people live and how they travel by road and air have been included. Our team has also gathered remote sensing data and ground observations on environmental factors that are important to malaria transmission, such as temperature, rainfall, and elevation.

By consolidating so many aspects of malarial dynamics, the model can make specific projections about which combination of eradication measures will have the greatest likelihood of suc­cess in a region, given the particular geography, climate, season, and environmental and social conditions that are in place there.

Our model also takes cost into account. For the first time, it allows campaign planners to weigh the inevitable trade-offs between the price of a malaria suppression strategy and its likely outcome. In short, it will show which malaria-fighting measures should be taken in which countries to ensure the greatest prob­ability of eradication at the lowest possible cost. It will thus help stretch antimalaria aid dollars to save more lives.

Preliminary runs of the simulator on our supercomputer cluster have been extremely promising. In just a few hours, the soft­ware is able to create a day-by-day simulation that spans years and covers an entire country such as Madagascar. For each day, prevalence of the disease is projected with very high spatial resolution. But the output is much more extensive than just prevalence. The model also produces daily projections for the population of mosquitoes, the number of malaria cases likely to receive clinical diagnoses, the number of lives claimed, the num­ber of mosquitoes infected with the parasites, and other crucial variables that will make it easier to validate whether the model accurately captures the most important aspects of the disease.

From the outset, the software was written in a modular way so that aspects of the biology of the parasite are kept separate from those of the insect that spreads it, and the code that handles the population is distinct from the procedures that process envi­ronmental variables. This highly modular architecture makes the model very flexible and relatively easy to adapt to simulate other infectious diseases. By tweaking the model in a few crucial ways, researchers will be able to use its core to simulate the dynamics of essentially any other pathogen. Ultimately, we hope this effort will produce a universal software tool for planning and executing infectious disease eradication campaigns.

A Persistent Pandemic

Scientists have understood the basic biology of malaria for more than a century. Richer parts of the world were able to eliminate malaria by waging lengthy, expensive campaigns that treated people with the antimalarial drug chloroquine, killed mosquitoes by spraying DDT, and drained the wetlands where the insects breed. But DDT is rarely used these days, and malaria parasites have developed resistance to chloroquine in many areas.

In poorer parts of the globe, where health care services are few and far between, malaria persists and exacts a huge human and economic toll. It is a leading cause of death and disease in many developing countries, sickening roughly one-quarter billion people each year and killing more than one million. According to the World Health Organization, in countries with high levels of malaria transmission, the effects of the disease decrease annual gross domestic product by as much as 1.3%.

Today, officials employ a mishmash of approaches to combat malaria transmission. Besides pesticides and antipara­sitic drugs, malaria-control programs install window screens and bed nets, drain standing water, and regulate land use. But whether a particular tool is effective in a given locale depends on many factors.

Different species of mosquitoes transmit the parasite in different locations, for example. Each species has a slightly different life cycle and set of characteristic behaviors. Drug re­sistance to chloroquine and other antimalarials has emerged as a major problem in some regions, but not all. Local geography and weather also affect, sometimes dramatically, how quickly the disease is transmitted. With so many variables interacting in non-obvious ways, it can be staggeringly complex to figure out the best combination of approaches for any given spot— especially when programs need to be inexpensive in order to be realistic.

The software that Dr. Eckhoff’s modeling team at the Intel­lectual Ventures Laboratory has created solves this thorny problem. Wherever possible, the simulation is built up from first principles, incorporating accumulated scientific knowledge and observational data. Mosquitoes, for example, are complex objects in the model with realistic behaviors. In the model, vir­tual mosquitoes feed outdoors or indoors according to known preferences for their species. They similarly breed at realistic (and different) rates in puddles and in large bodies of water, or prefer the countryside or the city as appropriate for their vari­ety. The model even tracks whether they typically land on walls and so might be susceptible to insecticide sprayed there.

Actual patterns of drug resistance in different areas of the world are not only incorporated, but are updated as waves of people migrate from one place to another in the virtual world. The software even includes a module that simulates how the parasites in each virtual person attempt to evade their host’s immune system by periodically changing the protein molecules adorning its surface—and how the antibodies of virtual deni­zens adjust in response.

Exploring All Possibilities

The tremendous detail of inputs and interactions that go into the model are one of its principle strengths. But the approach we have taken is remarkable in another way. Epidemio­logical models usually represent an infectious disease rather simply as a set of continuous equations, which the modelers tweak to fit historical data. That approach is good for forecast­ing the past, but has a checkered record when it comes to projecting the future.

A better approach, often used in the physical sciences, is called Monte Carlo modeling. Here the math is based on many relatively simple interactions, and it takes into account distinct states that occur, for example, in the life cycle of the parasite. More important, the results of the model are not a single calculation, but a statisti­cal blend of the results from running the model thousands or even tens of thousands of times, with a slightly different set of input scenarios each time. Because there is no way to know exactly what the weather or migration or mosquito population is at any given moment, this approach computes what would happen for myriad possible futures. It gives a more accurate projection of whether a particular campaign is likely to suc­ceed or in driving the number of infections to zero.

Simulations may show, for example, that installing bed nets in 75% of homes will reduce mosquito bites by too little to extirpate the disease, whereas 80% coverage will likely achieve eradication. The model could show that 20% more rainfall than expected is likely to cause a planned campaign to fail. It could show that the best time to spray insecticide on walls in a certain region is July rather than May.

Crucially, the model also totals up the cost of each eradi­cation strategy it evaluates. The simulations thus can opti­mize campaigns to achieve the greatest number of infections prevented—or lives saved—per dollar spent. We aim to even­tually be able to provide any malaria-affected region with an array of possible eradication plans, along with good estimates of the cost and probability of success of each plan.

We are currently improving this working model of malaria transmission by testing it against data collected both in laboratories and in the field. Although most of the data used to construct and test the model comes from other sources, we also perform some of our own lab experiments to fill in knowl­edge gaps and to make the model even more robust.

Although the software tool can be readily expanded to simulate infectious diseases beyond malaria, not enough labo­ratory and epidemiological data is not available in some cases to create a truly reliable eradication model. Even for these diseases, however, a working model can help prioritize future research. Simulations can, for example, reveal which aspect of a disease’s transmission is likely to have the largest influence on our ability to eradicate it. As a tool for both greater under­standing and more effective action, our infectious disease model is an exciting step toward a healthier humanity.

Download this white paper.