Researchers from the Universitat Oberta de Catalunya (UOC) they have created a technology based on neural networks to identify disease-carrying mosquitoes, like the tiger mosquito. The team belongs to Scene Understanding and Artificial Intelligence Lab (SUnAI) of the Computer Science, Multimedia and Telecommunications Studies and the eHealth Center of the UOC.
In the study they worked with tiger mosquitoes, invasive in nature in Spain and which can be carriers of dengue or Zika. Gereziher Adhane and Mohammad Mahdi, co-lead authors of the research, they have explained to Efe that, at first, they proposed “analyze images that were captured outside of laboratory conditions”.
To do this, the researchers used images that had been uploaded to the platform Mosquito Alert. This space was launched in 2014 with the aim of monitoring and controlling disease-carrying mosquitoes.
According to Adhane and Mahdi, thanks to the Mosquito Alert photographs, they were able to identify different types of mosquitoes. This task was, according to them, “important and critical”, Since insects that are disease carriers “pose a significant concern”.
The researchers claim that manual identification “expensive, time consuming, and nearly impossible under conditions other than laboratory conditions”. For this reason, the technology proposed by his study represents an important advance.
One of the best results of this identification system has been carried out with the tiger mosquito. This species is spreading more and more in Spain, which is worrying, since it can carry diseases such as dengue or Zika. With regard to the tiger mosquito, Adhane and Mahdi claim to have achieved “90% sorting accuracy”.
In addition to this insect, the researchers explain that with their automated approaches they could identify other species and monitor their large-scale expansion “in a short period of time”.
What will they do with this technology?
Adhane and Mahdi have not yet applied their system, however, they have their proposals: “Using this strategy, a mobile application capable of recognizing mosquito species could be developed.”. With this measure, the researchers mention that any user could upload a photo of a mosquito to the app and obtain information about the species.
With that possible platform, the expansion of said mosquito could be tracked and possible outbreaks of diseases that it may be a carrier.
As already mentioned, this identification technology works with deep neural networks. They go through a learning process: “This feedback is used to reduce the difference between actual values and those that have been predicted.”.
This system proposed by Adhane and Mahdi could be used by any individual who had the app installed on their mobile device. In this way, the presence of certain mosquitoes could be reported to curb their invasive nature in early stages and monitor outbreaks of some diseases.
Sign up for our newsletter and receive the latest technology news in your email.