App can tell you if a mosquito is about to give you malaria

A mosquito bite can infect you with malaria, dengue or zika, diseases that kill hundreds of thousands of people every year. But how can you tell if the mosquitoes near you are dangerous? Well, now there’s an app for that. Only 40 of the 3500 mosquito species bite humans. It would be nearly impossible to identify the dangerous kinds by sight alone: two species of mosquito may look identical, for example, but one prefers animals to humans. However, every species has a unique sound signature. They “sing” to identify each other in mixed-species mating swarms, says Marianne Sinka at Oxford. To take advantage of this, teams at both Oxford and Stanford are using the surprisingly sensitive microphones on smartphones to record a mosquito’s whining wing beat and match it against a database of known examples for identification. Oxford’s “MozzWear” Android app lets researchers use the phone’s microphone to record a mosquito’s buzz from up to 10 centimetres away. Even the most basic £25 mobile phone will do the job. The app then compares the acoustic profile using a bespoke machine-learning algorithm. In pilot trials, accuracy rates across seven species range from 68 to 92 per cent. Stanford’s Abuzz system lets anyone record and upload mosquito sounds to a website, which uses an algorithm to match and identify them. They are working on a version that works over text messaging: you would send the buzz as a voice memo, and find out what type of mosquito it was via SMS reply.

More than an app

The Stanford project can identify 20 species so far, with accuracy ranging between 70 and 90 per cent if other data from the phone, such as time and location, are included, said Manu Prakash, who is leading the Stanford work. The apps will do more than just help researchers map mosquitos, says Heather Ferguson of the University of Glasgow. Local authorities and communities affected by malaria or zika can use them to decide where to spray to prevent disease. These may also act as an early warning system for new species entering an area. While the results and pilot trials are promising however, they still need to show they work “under a fuller range of ecological conditions”. That includes background noise as well as variations in mosquito traits within a species. Larger body sizes influence wing beat, Ferguson said, which could lead to false negatives. Sinka will present the work at the Conference on Neural Information Processing Systems in December.

New Scientist, 23 November 2017 ; http://www.newscientist.com/