Alphabet Is Using AI To Help Rid Communities Of Disease-Carrying Mosquitoes

If all goes as planned, Alphabet’s life-sciences unit is about to help communities wipe out a whole lot of mosquitos.

Verily, the health company that sits under the Alphabet umbrella, has been quietly working for a few years on an effort known as the “Debug Project.” It is helping communities in the United States and beyond reduce populations of an invasive species of mosquito called Aedes aegypti. This particular species is a known carrier of potentially-fatal diseases like dengue, Zika, chikungunya, and yellow fever.

Verily’s steadily-growing team of mosquito biologists and computer scientists is focusing its energies on the “sterile insect technique,” which has been used for many decades to control mosquito populations. The idea is to release infertile male insects into the wild to mate with females, that will then release eggs that won’t hatch. It is most effective in controlling mosquito species that tend not to travel much in their lifetimes, and that only mate once. On both those counts, Aedes aegypti is an ideal candidate.

This might seem like an odd proposition for a company that was formerly part of Google X, but Linus Upson, Verily’s vice president of engineering, says he has been interested in tackling this problem since his undergraduate days at Princeton University. “That’s when I found out how many people that mosquitos kill and sicken,” he says. “Cars kill more people than mosquitos by just a bit, but others [at Alphabet] are working on that problem.”

Alphabet is also the parent company of Google, which is working on bringing autonomous vehicles to the road.

In recent years, proposals that would genetically modify insects in order to control populations and fight disease have proven controversial. In Florida’s Key West, where Zika cases have been reported, a biotech company called Oxitec unveiled a plan to inject mosquito eggs with DNA that contains lethal genes. Many residents shared concerns about the plan, including fears that they were being used as human experiments.

For this reason Verily’s Debug Project is focusing on “non-GMO” alternatives: Sterilize the male mosquitos by infecting them with a naturally occurring bacteria called Wolbachia. It has been known for some time that this bacteria, which is possibly the most common reproductive parasite, will cause the males to become sterile while leaving them physically up to the task of competing for females in the wild. Earlier this summer, a company called MosquitoMate filed for approval from the Environmental Protection Agency to get Wolbachia in a related mosquito, Aedes albopictus, as a pesticide.

Some labs, such as Stephen Dobson’s at the University of Kentucky, have already tested the Wolbachia technique in the field in relatively uninhabited places. And in August, a city called Clovis in California released 400,000 males carrying Wolbachia into the wild to control Aedes aegypti.

What Verily is bringing to the table is its machine learning and computer vision expertise. The company is developing algorithms to rapidly distinguish between the male and female mosquitos by picking up on subtle physical characteristics that are almost impossible to detect with the naked eye.

In previous field tests, researchers would often have to hand-separate the insects. That can be costly and result in mistakes. It’s important to avoid releasing females, as they’re are the ones that rely on blood meals while the males subsist on plant nectar.

In a blog post, the company shared that it is also working on prototypes related to automated rearing and releasing the insects, as well as new sensors to track the mosquito population.

“Any way that you can automate and increase accuracy without accidentally releasing females is good,” says Jason L. Rasgon, an associate professor of entomology and disease epidemiology at Penn State, when asked about the role that Verily might play. If the company can show the efficacy of its technology in peer-reviewed papers, Rasgon says it might lead to this Wolbachia technique being “more widely adopted.”

At this stage, Upson admits that the technology is far from perfect, but that it’s getting smarter over time. “We’re at the prototype phase right now,” he says. “But we can achieve a high level of accuracy.”

In recent months, Rasgon says he’s seen the company recently pluck some of the “excellent minds” out of academia, including fellow entomologists, to work on this problem. Verily’s Upson wouldn’t share how many experts it has hired to date. “Let’s just say that we’re making a significant investment.”