We tell what it is from its buzz – Artificial Intelligence in mosquito monitoring

Mosquitoes are responsible for transmitting several serious diseases, including malaria, dengue fever, chikungunya, and Zika fever. Together, these diseases affect millions of people annually and cause hundreds of thousands of deaths worldwide. The most effective form of protection is prevention, which requires continuous monitoring of mosquito populations. This enables early detection of risks and timely implementation of necessary interventions, such as mosquito control measures.

During flight, mosquitoes produce sound through the beating of their wings: the faster the wingbeats, the higher the frequency of the sound. This acoustic signal varies by species, which is particularly useful because it allows monitoring efforts to focus only on species that pose actual risks, such as disease-carrying or invasive mosquitoes.

AI-based algorithms that can identify mosquito species based on sound with up to 97% accuracy already exist. However, the method has limitations—for example, when multiple species are present simultaneously, when there is insufficient audio data available for training, or when the sounds of wild mosquitoes differ from those recorded under laboratory conditions. Acoustic signals are influenced by numerous factors, including temperature, humidity, sex, age, and individual body size.

This is why researchers from the HUN-REN Centre for Ecological Research, Eötvös Loránd University, and the University of Szeged conducted a detailed investigation into how these environmental and biological factors affect the variability of mosquito sounds. They analysed the acoustic data of 475 individuals from ten mosquito species found in Hungary and concluded that these sounds carry information not only on the species level, but also about individuals. The accuracy of species identification can therefore be further improved if AI applications take these variables into account.

For example, it was found that female mosquitoes generally produce lower-frequency sounds than males, which is related to their larger body size. Temperature also plays a key role: in warmer environments, mosquito muscles move faster, increasing the number of wingbeats per time unit and resulting in higher-frequency sounds. However, this effect varies by species, meaning that a single temperature correction cannot be universally applied.

The findings of this recent study represent an important step toward enabling the effective application of artificial intelligence under real-world conditions for monitoring hazardous mosquito species.