Mobile phones identify mosquitoes
Chris - As the vectors that transmit malaria, dengue, yellow fever, Zika and a host of other infections, mosquitoes are widely regarded as one of the world's most dangerous animals. Yet we know very little about where they are, how they are spreading, and how their distributions influence disease. Now mobile phones may be coming to the rescue. From Stanford University, Manu Prakash...
Manu - You know as many problems, they start with experience and many of us have experience being bitten to bits by mosquitoes. I was in a field trip in Thailand when I met a lot of medical entomologists hunched back on microscopes essentially sorting mosquito species and made a realization that today we actually use the same technology that we were using hundred, 150 years ago to speciate and identify mosquitoes in the field. So the problem has always been is we know mosquitoes are important. I don't have to repeat the number of deaths caused by them with all the diseases and all the implications that we don't actually know for these diseases but we don't know where our enemy is. And that's what we wanted to solve. Could we imagine a new platform that is scalable, very very low cost, and it provides a new window essentially into the life of mosquitoes?
Chris - Why does it matter? Identifying what type of mosquito you're dealing with.
Manu - There are more than thirty-five hundred species of mosquitoes and only a few of them, between 20 to 40 different species are the ones that carry deadly diseases. For many of these diseases, we actually don't even have a cure. So it becomes extremely important to be able to understand where these vectors are. Both for scientists who are trying to understand these diseases but also for municipalities who are trying to reduce the burden of these diseases by reducing mosquito populations
Chris - Because those sorts of authorities have been tracking where the mosquitoes are and in what sorts of numbers by doing things as primitive as just putting out a trap and then counting what they catch haven't they?
Manu - Yeah and you know it's a very effective strategy it's just you need a lot of people to do that. So it doesn't scale to the size of our planet and specially with many of these mosquitoes even spreading their range where they're found. So we need a better solution.
Chris - So what is your solution?
Manu - We've been thinking about this problem for awhile and we stumbled upon a very old known fact which is the fact that mosquitoes fly they generate this buzzing sound and we made a realization that using a regular cell phone you can record these sounds and using machine learning and algorithms we can actually speciate these recordings into specific mosquito species.
Chris - So there is a signature wing beat that goes with a species is there? And you can use that to discriminate them?
Manu - That's correct. And both males and females have different wing beat frequencies and because frequency and sound plays an integral role in how mosquitoes mate they actually play love songs to each other, in fact. It essentially ends up happening that many species have differentiated wing beat sound signatures and with cell phones, what's powerful is you don't just get the acoustic signature, you actually get the location of where that recording was made, the time of the recording. This is all served as metadata and using that metadata we can even improve the accuracy of the classification to be even higher.
Chris - A mobile phone is good enough is it the microphone that's in there is adequate to capture a clean enough signal of the sound in order for you to get the diagnosis of what the mosquito is?
Manu - Yeah and this was a really fun finding. We compared almost five to six different brands of cell phones at different prices and what we demonstrated in the paper is it turns out because cell phones are designed to record and transmit sound, the microphones have been improving over the years to an extent that they are actually fantastic both for capturing the sound and transmitting it and recording the meta data and we demonstrate that they actually compete very well with some very expensive microphones as well to do the job.
Chris - So have you done field trials on this? Have you now got citizen scientists, people out, stakeholders and actually demonstrated that this can accurately tell you what sorts of mosquitoes are buzzing around those individuals at that point in tine?
Manu - Yeah that's always been why we started doing this to begin with is to engage the community and ironically you don't have to be in the bushes to do this because many mosquitoes live in urban environments. When we wrote the paper there is a website that anybody can go to. We've gotten lots and lots of recordings coming from around the world from rural India to Zaqistan to rural parts of United States, all across the world. And one of the big things that we're trying to do is make the tools easier and easier for the community to engage and specially the new sets of tools that we're building on top of this technology is things that provide them immediate feedback, provides them more information about their local surroundings and the types of mosquitoes that they're recording. So this is ongoing piece of work. You can go to our website "abuzz dot stanford dot edu" and you can subscribe and also upload your recordings and very soon we will be releasing an app. Which will allow to integrate this directly in the cell phones that people use.