The researchers at the University of Wollongong (UOW) India have developed a low-cost, artificial intelligence (AI)-powered device capable of identifying disease-carrying mosquito species within seconds by analysing the sound of their wingbeats.

The innovation comes at a time when mosquito-borne diseases such as dengue, malaria and chikungunya continue to pose a major public health challenge in India, particularly during the monsoon season when outbreaks typically surge across several states.

Developed by Associate Professor Kiran Trivedi in collaboration with former student Harsh Shroff, the portable device leverages Tiny Machine Learning (TinyML) to identify three of the world's most important disease vectors—Aedes, Anopheles and Culex mosquitoes—without requiring internet connectivity or cloud computing.

Unlike conventional mosquito surveillance methods that rely on collecting larvae and laboratory-based species identification, the AI-powered system recognizes mosquitoes by analysing the unique acoustic signatures produced by their wingbeats. 

The embedded AI model processes sound locally on the device, enabling real-time identification while operating on minimal power.

Built on a compact Arduino-based platform equipped with an integrated microphone and display, the device offers a portable and cost-effective solution for field surveillance. The AI model, trained using publicly available mosquito sound datasets, demonstrated an identification accuracy of 88.3%.

The innovation was recently showcased at the United Nations AI for Good Global Summit in Geneva, where Associate Professor Trivedi demonstrated how affordable edge AI technologies can be deployed to address critical public health challenges.

Associate Professor Kiran Trivedi, University of Wollongong India, said: "TinyML allows us to bring artificial intelligence directly onto a small, portable device, enabling mosquito species to be identified within seconds without internet connectivity or laboratory infrastructure."

Highlighting its broader public health potential, he added: "Our objective is to demonstrate how low-cost, accessible AI solutions can be applied to address pressing public health challenges, particularly in regions where traditional surveillance infrastructure may be limited."