A machine learning approach to enhance mosquito repellent effectiveness (2025)

In a recent study, researcher Anandasankar Ray at the University of California, Riverside, and his team employed machine learning techniques combined with cheminformatics to predict novel mosquito repellents that could greatly improve global mosquito control efforts.

The findings are published in the journal eLife.

Using the same approach to combat the global threat of mosquito-borne diseases such as malaria and dengue, Ray will work on identifying novel spatial mosquito repellents and their mechanisms.

Mosquitoes use their olfactory (smell) and gustatory (taste) systems to detect and feed on human hosts. Current skin-applied insect repellents, such as DEET, are effective but costly, need frequent reapplication, and suffer from poor user experience, especially in low-income tropical regions.

Spatial repellents, which emit pyrethroid insecticides (synthetic insecticides for killing mosquitoes) in low doses through heated dispensers or coils, are widely used, but are facing rising concerns due to the rapid spread of mosquito resistance to pyrethroids.

The machine learning-based cheminformatics approach Ray's team developed has screened more than 10 million compounds for potential new repellents and insecticides. Using this approach, Ray and his team have identified novel repellent compounds from natural sources, such as common food and flavoring materials, that are effective and pleasant-smelling.

"We have already identified repellent molecules with a high success rate, particularly from natural sources, which could provide a safer and more sustainable alternative to current repellents," said Ray, a professor of molecular, cell and systems biology.

"We have also used machine learning to identify analogs of pyrethroids that are up to 100 times more effective than existing industry standards, like allethrin. This could have a significant impact on combating resistant mosquito populations."

The proposed research aims to identify the most effective insect control compounds across four key categories.

These are improved topical repellents, which provide long-lasting, pleasant-smelling protection for over 12–24 hours; spatial repellents designed to protect spaces such as backyards and houses from mosquitoes; long-lasting pyrethroid analogs, which are new pyrethroid-like molecules effective against resistant mosquito strains, making them ideal for use in bed nets and clothing; and enhanced spatial pyrethroid formulations, which offer increased effectiveness against mosquitoes exhibiting knockdown resistance (resistance to pyrethroid insecticides).

Ray's team will also use mosquito mutants to pinpoint the receptor pathways responsible for aversion to new repellents. The researchers will test volatile compounds for spatial protection and evaluate new pyrethroid analogs for efficacy against resistant mosquito strains.

"By identifying and combining the most effective natural and synthetic compounds, we hope to deliver safe, affordable, and highly effective mosquito control solutions that could help reduce human exposure to disease vectors while improving quality of life in at-risk populations," Ray said.

"We are looking for repellents that work as well as cost-effective, easy-to-use, and culturally acceptable solutions. Based on our preliminary results, we are optimistic that the new compounds could soon be a new weapon in the fight against mosquito-borne diseases."

A matter of taste

Ray is the principal investigator in other work aiming to understand why some humans are less attractive than others to mosquitoes.

Mosquitoes use their sense of smell and taste to find and feed on humans, spreading diseases like dengue. These sensory systems are key targets for developing better repellents. The gustatory system, which helps mosquitoes avoid DEET, has not been fully explored.

"There's a need for more effective repellents since DEET's high cost and poor properties limit its use in tropical areas," Ray said. "We believe compounds in human skin, sweat, and microbiome metabolites could be key."

According to Ray, the project aims to identify skin compounds that influence mosquito landing behavior and analyze the chemoreceptor pathways involved.

"We will test these compounds in behavior assays, focus on those that affect taste or smell, and explore how blends of repellents may reduce mosquito attraction," Ray said. This research could lead to improved mosquito control strategies.

Ray's team will collaborate on both projects with Anupama Dahanukar, a professor of molecular, cell and systems biology at UCR.

Previous work from Ray's lab led to a product development plan by the National Institute of Allergy and Infectious Diseases and a university spinoff company, Sensorygen, that has led to a safe and natural lead insect repellent being evaluated for registration at the Environmental Protection Agency.