Bandits On The Loose: Combatting Chagas Disease Through The Power Of Data

​While the mosquito-borne Zika virus continues to dominate news headlines around the world, another critter is causing similarly troubling health concerns throughout South America.

While the mosquito-borne Zika virus continues to dominate news headlines around the world, another critter is causing similarly troubling health concerns throughout South America.

It’s called Chagas disease, which is also known as the “kissing bug” because it leaves painless bites on a victim’s skin. But the blood parasite, officially classified as Typanasoma cruzi, is much more insidious than its nickname implies: Known to attack victims in their sleep, kissing bugs spread Chagas disease.

See why the World Economic Forum recognizes Uptake for revolutionizing the construction industry

The parasite remains in the victim’s body for decades, causing gradual damage to tissues such as the heart and the digestive system, and in many cases, death. Researchers have estimated that the disease kills more people in South America than malaria.

This parasitic disease resides mostly in South America, though it has established a small foothold in the American southwest. The Pan American Health Organization estimates there are about 8 million cases in Latin America, while the CDC estimates about 300,000 people are infected in the United States.

Uptake data scientist Sasha Gutfraind has been studying the public health implications of Chagas disease for several years. Recently, he took some time off from his day job at Uptake to return to Arequipa, Peru, where he’s been logging time at a lab with a team of researchers working on identifying more efficient ways to stop the spread of Chagas in dense urban residential areas like Arequipa, which is Peru’s second largest city.

Putting data science to work on this vexing public health challenge, the power of data can enable new approaches to some of the world’s hardest problems, and it is representative of the diverse studies Uptake’s data scientists pursue as part of an overarching strategy to better inform how we build solutions across industries.

Because there is no vaccine for Chagas disease, the most effective preventative method is a process known as vector control: Infected areas are treated with insecticides, with crews going house-by-house. Follow-up surveillance then is conducted to ensure infestations are under control. Yet, in urban areas like Arequipa, spraying house by house and then conducting follow ups can be an impractical, cost-prohibitive solution—and therefore rarely implemented by local authorities.

A shanty town in Arequipa, Peru. Low quality of construction and yard animals provide a habitat for kissing bugs that spread Chagas.
A shanty town in Arequipa, Peru. Low quality of construction and yard animals provide a habitat for kissing bugs that spread Chagas.

Sasha and his research team developed algorithms for more efficiently finding infestations using online optimization, which used the method of multi-armed bandit (MAB) algorithms to identify search areas for Chagas infestations in Arequipa and accelerate the process. With this method, the research team’s field survey group reports its results a daily basis, which is used by the MAB algorithm to plan work for the next day.

This is the first application of online optimization for insect vector control. Data collection is ongoing, with early results expected by the end of the year. In the meantime, the team is performing computational simulations on real data to provide a better understanding of the limitations of the algorithm.

These types of experiments are reflective of the diversity of projects that Uptake data scientists undertake to better inform their work for our partners. Rather than be limited to one specific sector or structure, Uptake’s data science team looks at data across seemingly disparate areas to formulate a more informed outlook. Having that more universal understanding of how data works in one segment versus another, it enables Uptake’s platform to create ever more accurate predictions as we integrate findings discovered from patterns in data across a wide array of applications.