Jakarta, INTI - Associate Professor Kiran Trivedi from the University of Wollongong (UOW) in India developed a low-cost portable device that can identify disease-carrying mosquitoes by the sound of their wingbeats. This technology offers a faster surveillance method than conventional methods for monitoring the spread of diseases such as malaria and dengue.
The device is built on artificial intelligence (AI) Tiny Machine Learning (TinyML), and it can quickly identify three of the world’s most significant disease-carrying mosquito species, namely Aedes, Anopheles, and Culex.
The device can be used without internet connection. The TinyML system allows AI models to run directly on small low-power chips rather than relying on high-capacity computers or cloud systems.
For this innovation, Trivedi was invited to demonstrate the technology at the United Nations AI for Good Global Summit in Geneva.
According to the World Health Organization, mosquitoes are the world's deadliest animals, causing hundreds of thousands of deaths each year. The heaviest impact is felt in developing countries and remote communities with limited access to laboratory facilities.
Faster Monitoring Process
Traditional mosquito monitoring is usually conducted by taking water samples from breeding sites and analyzing larvae in the laboratory to determine the species. Trivedi found that wingbeat sounds could be a faster monitoring alternative, as each mosquito species has distinct wingbeat patterns and produces a subtly distinct acoustic fingerprint.
“When people think about AI, they imagine huge systems running in the cloud. TinyML lets us put the intelligence directly onto the device. It identifies the mosquito in seconds, with no internet, no cloud costs and no privacy concerns,” said Trivedi.
The AI model was trained using publicly available audio recordings and achieved an accuracy rate of 88.3%. Trivedi considered the result to be quite good for audio classification and has the potential to be improved through the use of better microphones and cleaner recordings. The system runs on a small Arduino-based device equipped with a microphone and a display.
Associate Professor Trivedi believes the technology’s real power lies in scale, with networks of devices monitoring mosquito activity around the clock and feeding results into live maps.
“Just as a navigation app shows you traffic in real time, this could show where disease-carrying mosquitoes are building up. Instead of waiting for an outbreak, communities and public health agencies could see the hotspots early and respond,” Trivedi explained.
The research, co-authored with his then-student Harsh Shroff, was first published in 2021 in the International Telecommunication Union’s Kaleidoscope conference proceedings.
Conclusion
Associate Professor Kiran Trivedi from UOW in India has developed a low-cost, portable device that uses AI TinyML to identify disease-carrying mosquitoes based on the sound of their wingbeats. The device can identify Aedes, Anopheles, and Culex species in seconds without an internet connection. The system offers a faster alternative to traditional surveillance methods that require sampling and analyzing larvae in a laboratory, especially for developing countries and remote communities.
Read more: Indosat Assesses How AI and IoT Can Be Utilized for Energy Management