A Microsoft research team is partnering with Harris County in Houston, Texas, to tests a new type of mosquito trap. The initiative, called Project Premonition, leverages new technology to trap and test mosquitos for diseases like Zika and Dengue fever.
The trap collects data on the time and weather conditions when each mosquito is collected. The device only traps certain types of mosquitos so that entomologists don’t have to sort through a variety of insect species. It uses two battery-powered microprocessors to gather information and transmit wirelessly to the cloud. Using machine learning, the trap can analyze the data and select the type of mosquitos researchers are interested in. The trap is also weatherproof, allowing it to hold up in storms where other devices might break down.
When Mustapha Debboun, Director of Harris County Public Heath’s mosquito control division, heard about the technology, he reached out to Microsoft asking to get involved. The Redmond-based software giant agreed to test the trap in the region, which is too large to effectively combat mosquito-born diseases with insecticide.
“For a scientist this a dream come true,” Debboun said in a Microsoft story on the announcement. “This is something so new, so innovative, so technologically advanced.”
Microsoft describes the trap as an “early warning system” to identify diseases like Zika before patients start turning up in hospitals. The goal is to trap mosquitos in rural areas and catch any potential health concerns before they reach urban regions.
Project Premonition originally planned to develop public health technology over five years but decided to launch early, due to the imminent Zika threat.
Eventually, Project Premonition will use drones to observe mosquito habitats and more efficiently place the traps in remote areas.
Testing the traps in Harris County will help Microsoft collect the data it needs to hone the technology. The device identifies which kinds of mosquitos to trap based on the way the insects flap their wings. In order to perfect that sorting system, the system’s algorithm needs to observe examples of bugs. Through machine learning, it will become smarter at identifying the type of mosquitos researchers are interested in over time. In the short term, the trap will help areas like Harris County identify mosquitos that pose a threat and curb the spread of diseases they bear.
Debboun describes the new trap as “really 1,000 times better” than other early warning systems.