Engineers, computer scientists, virologists, and epidemiologists team up to nip the next mosquito-borne epidemic in the bud.
Imagine if scientists could detect mosquito-borne infectious diseases in the environment before they cause potentially deadly outbreaks of Zika virus, West Nile virus, dengue fever, or other dangerous diseases.
Enter the annoying and sometimes dangerous mosquito, which also doubles down as a helpful device.
“Mosquitoes are the ultimate field biologists, taking blood samples from every animal they bite,” said Ethan Jackson, who works in the Research in Software Engineering (RiSE) Group at Microsoft Research in Redmond, Washington.
Jackson is leading Project Premonition, a groundbreaking endeavor that includes scientists and engineers at the University of Pittsburgh; Johns Hopkins University; University of California, Riverside; and Vanderbilt University, as well as Harris County Public Health in Houston and other public health agencies.
The project brings together computer scientists, engineers, virologists, epidemiologists, and entomologists. Together they're pushing the boundaries of robotics, machine learning, and metagenomics to nip the next mosquito-borne epidemic in the bud.
Training the Trap
Project Premonition autonomously obtains blood samples from animal populations via robotically collected wild-caught mosquitoes. Mechanical smart traps lure the troublesome insects by wafting carbon dioxide their way. Prototype versions of the traps then collect the insects to return blood samples and body contents that reveal the animals, bacteria, and viruses they’ve consumed.
The researchers use next-gen, high-throughput DNA sequencing to convert samples into metagenomic data, then they use Microsoft Cloud to do computational genomics analysis, which identifies potential threats. This can spot diseases that infect both mosquitoes and their host animals, including humans. It can also spot previously unknown viruses.
“We want to understand the possible pathogens mosquitoes encounter in the environment to stop outbreaks before they begin,” Jackson said.
The ambitious project debuted last summer in Harris County, Texas, the third most populous county in the nation and home of the fourth-largest city, Houston. It included 87 experiments that captured nearly 20 GB of data about mosquitoes’ most private moments.
“The data that‘s collected allows a public health system to fine-tune its responses and consider actionable decisions such as, ‘Do we spray more?’ or ‘Do we need to use trucks?’” Jackson said. “That sort of data has been difficult to obtain and it’s pretty coarse. Our system provides much finer-grained data that allows public health organizations and the rest of us to make better decisions.”
Fly by Night and by Day
Last summer’s Texas outings did not include drones, but this summer’s will. Test runs of the improved traps were conducted on the Caribbean island nation of Grenada. Currently, the teams use the 10-inch tall, 18-inch long 3DR Solo drone, which can fly up to 0.6 miles at up to 55 mph. At some point, modified versions of the drones may actually fly traps in and place them, saving humans the task of getting there on two feet, the researchers said.
This could save a lot of time and make for more comprehensive data collection, since Harris County is 1,777 square miles, and there are only so many public health workers to place traps. “In the near future, we’d like to test up to 400 new Microsoft smart traps at once,” said Mustapha Debboun, who directs mosquito and vector control for Harris County Public Health. “I’ll use as many as Ethan can provide me for the 268 mosquito-operational designated areas we sample.”
DIY Traps Coming
The prototype traps the Project Premonition scientists used required them to sift through as many as 5,000 insects—mostly mosquitoes. The new traps selectively trap only mosquitoes that could can carry disease. “We ‘train’ the new traps to actually ‘screen’ for us,” Jackson said, “and that’s so important in public health. We want to concentrate on mosquitoes that carry disease to know where they are and possibly forecast for disease in that area.”
The traps include 64 smart cells. To trap only the desired mosquito species, each cell uses infrared light and sensors to detect the frequency of a mosquito’s beating wings. Data captured by those sensors then trains and improves the trap's control algorithm. Before long the smart cells learn to recognize when mosquitoes of a targeted species fly in.
The trap also collects data on temperature, relative humidity, barometric pressure and the amount of light. Two battery-powered microprocessors do the hard work, assimilate the data and send it to the cloud via wireless technology.
“We want to know right away when they’re trapped, when they fed last and on what,” Debboun said. He looks forward to the day—perhaps in three to five years—when all of us can have a robotic trap in our yard, at an affordable price, along with fewer itchy mosquito bites.
Stephanie Stephens produces health content in Orange County, California.