Photo courtesy of USF
By Jasmin Parrado
Researchers at the University of South Florida announced their goal to patent a smart trap that can use artificial intelligence (AI) to identify disease-carrying mosquitoes. They believe the trap can aid malaria research and prevention in vulnerable communities.
The smart trap, which is trained on algorithms of compiled image data, is able to identify the sex and species of any mosquito it captures, a surveillance power that entomologists believe is crucial to understanding a mosquito’s disease-carrying capacity.
Ryan Carney, co-creator of the smart trap and professor of digital science at USF’s Department of Integrative Biology, believes the trap is integral to protecting areas where disease-carriers, or vectors, roam, especially because of how efficient it can be in spotting disease risks early on.
“It’s basically a race against time where we need to detect it, especially when it’s in new areas, because you have about a three-to-four-year window before it becomes established,” Carney said. “The identification of these mosquitoes can be very challenging as well as time-consuming. So, by developing these algorithms, we can do that much faster.”
Out of 3,500 mosquito species, only 3% are vectors, Carney explained. With 14 vector species detected in Florida, epidemics like the ongoing Dengue outbreak in South Florida and the Zika virus that reached the state in the mid-2010s have led to national efforts to track mosquitoes.
A mosquito species that particularly interests researchers is Anopheles stephensi, the primary vector of malaria that is resistant to a number of environmental and man-made conditions.
Carney believes that his team was likely able to identify an Anopheles stephensi larval mosquito by training AI with an image database of anatomical characteristics in 2020.
“You can think of it as facial recognition for mosquitoes,” Carney said. “We had students both at USF, as well as the insectary at the CDC, take thousands upon thousands of images of larval mosquitoes. We then trained these algorithms to detect the various species.”
Sriram Chellappan, co-creator of the trap and professor at USF’s Department of Computer Science and Engineering, outlined the broader benefit of the trap’s algorithmic technology.
“When surveillance of vectors is fast, accurate, reliable and efficient, control efforts are naturally also fast and efficient,” Chellappan said. “This will prevent or mitigate disease spread.”
Carney and Chellappan’s smart trap is part of a larger initiative known as Enhancing Malaria Epidemiology Research through Genomics and Transnational Systems (EMERGENS) — a project funded by the National Institutes of Health.
The project also works through a public database Carney and Chellappan created in 2022, known as mosquitodashboard.org. Using reports from citizens of mosquito surveillance, the dashboard serves as an accessible visual on the presence or influx of mosquito species.
Photo courtesy of Mosquito Dashboard.
In addition to other forms of mosquito surveillance like smartphone applications and clip-on magnifiers, Carney and Chellappan are pushing for the smart trap to be accessible to citizens and researchers alike, in order to provide communities a chance to act quickly if disease-carrying species are present.
Chellappan noted that there are special considerations to explore when evaluating the trap’s technology.
“Are the AI models accurate? Is the data for training diverse enough? Can the model learn from new data sets without need to re-train? Speed of execution versus accuracy analysis are things of interest,” Chellappan said.
Carney explained how the project garnered interest from the government of Madagascar, leading to a collaboration with the CDC to enact a one-year mosquito surveillance program that revealed 1,300 Anopheles mosquitoes.
The finding was consistent with the outpour of citizen data that reached the database website, Carney recalled.
“By building that dashboard and bringing all the data together, local communities as well as mosquito control districts can see what mosquitoes have been detected in their area, in locations that are inaccessible and greatly scalable,” Carney said.
Carney and Chellappan are focused on monitoring malaria-carrying vectors, as the disease accounts for a large number of mosquito-related deaths in Africa. According to the World Health Organization, the region accounted for over 200 million malaria cases and more than half a million deaths in 2023.
“When you think of the world’s deadliest animals, you often think of something like a shark,” Carney said. “But they kill roughly a dozen people every year, and mosquitoes kill approximately 1 million people every year. 700 million infections, which is just staggering.”
