The Zika epidemic that emerged in Brazil in 2015 and left thousands of babies suffering from birth defects has added urgency to the effort.
While cases there have slowed markedly, mosquitoes capable of carrying the virus - Aedes aegypti and Aedes albopictus - are spreading in the Americas, including large swaths of the southern United States.
(For a map of U.S. mosquito territory, see http://tmsnrt.rs/2tqlJHa)
The vast majority of the 5,365 Zika cases reported in the United States so far are from travelers who contracted the virus elsewhere. Still, two states - Texas and Florida - have recorded cases transmitted by local mosquitoes, making them prime testing grounds for new technology.
In Texas, 10 mosquito traps made by Microsoft are operating in Harris County, which includes the city of Houston.
Roughly the size of large birdhouses, the devices use robotics, infrared sensors, machine learning and cloud computing to help health officials keep tabs on potential disease carriers.
Texas recorded six cases of local mosquito transmission of Zika in November and December of last year. Experts believe the actual number is likely higher because most infected people do not develop symptoms.
Pregnant women are at high risk because they can pass the virus to their fetuses, resulting in a variety of birth defects. Those include microcephaly, a condition in which infants are born with undersized skulls and brains. The World Health Organization declared Zika a global health emergency in February 2016.
Most conventional mosquito traps capture all comers - moths, flies, other mosquito varieties - leaving a pile of specimens for entomologists to sort through. The Microsoft machines differentiate insects by measuring a feature unique to each species: the shadows cast by their beating wings. When a trap detects an Aedes aegypti in one of its 64 chambers, the door slams shut.
The machine "makes a decision about whether to trap it," said Ethan Jackson, a Microsoft engineer who is developing the device.
The Houston tests, begun last summer, showed the traps could detect Aedes aegypti and other medically important mosquitoes with 85 percent accuracy, Jackson said.
The machines also record shadows made by other insects as well as environmental conditions such as temperature and humidity. The data can be used to build models to predict where and when mosquitoes are active.
Mustapha Debboun, director of Harris County's mosquito and vector control division, said the traps save time and give researchers more insight into mosquito behavior. "For science and research, this is a dream come true," he said.
The traps are prototypes now. But Microsoft's Jackson said the company eventually hopes to sell them for a few hundred dollars each, roughly the price
of conventional traps. The goal is to spur wide adoption, particularly in
developing countries, to detect potential epidemics before they start.
"What we hope is (the traps) will allow us to bring more precision to public
health," Jackson said.