Dr. Manu Prakash is once again at his best — solving complex problems in an extraordinarily simple fashion. Using wingbeat sound, location and time of recording, Dr. Prakash and his team from Standford University hope to make real-time global maps of mosquito distributions, which could help design better strategies to control mosquitoes and combat mosquito-borne diseases.
Using even low-cost mobile phones to record mosquito wingbeat sound while simultaneously recording the time and location and sharing the data with Abuzz project will help in global mosquito surveillance and control. The high-throughput, low-cost strategy to generate data has been developed by a team of researchers led by Dr. Manu Prakash from the Department of Bioengineering at Stanford University.
Dr. Prakash is famous for the cheap, foldable microscope called foldscope that he developed, which is widely used by many across the world.
When people across the world share wingbeat buzz data the team hopes to make real-time global maps of mosquito distributions, which could help design better strategies to control mosquitoes and combat mosquito-borne diseases. Once the recording (at least 1.2 second duration) of the wingbeat sound is shared with Abuzz website, the researchers will remove the background noise and run it through an algorithm that matches that particular buzz with the species that is most likely to have produced it.
According to the university release, the person who submitted the recording will be informed of the species and the team will mark every recording on a map on the website, showing exactly where and when that mosquito species was sighted.
The team used using various mobile phones to record the wingbeat buzz of 20 mosquito species that spread disease to humans. When the recordings were compared with collection of sounds, the correct species could be identified for 66% of the recordings in the case of a National Park in California and a village in Madagascar. The accuracy of species identification can be increased by simultaneously recording the location and time of the recording. The results were published in the journal eLife.
The wingbeat sound of mosquitoes has a natural frequency variation which varies in a range of characteristic for a particular species. Female wingbeat frequencies are lower than male mosquitoes and are in the range of 200 and 700 Hz. Though this frequency range overlaps with voice frequency, the team found that sound recorded by mobile phones even in a noisy environment can help in identifying the mosquito species.
The researchers recorded the wingbeat sound in the lab and field, and in ambient environments — urban and rural homes (both indoors and outdoors) and in forest and parks. “Wild mosquitoes in field environments may vary considerably in terms of age, body size, rearing temperature, and nutrition status, yet our method requires the variation in their wingbeat frequencies to be sufficiently small in order to be identifiable,” they write.
Using a decade-old phone, the team was able to record for distance up to 5 cm from the mosquito when the phone recorder was properly orientated. The sensitivity of recording is better in the case of a smartphone for distance up to 10 cm.
The team used eight different phones, both ordinary and smartphone, and found that the mean frequencies recorded by each phone varied only negligibly (less than 5%). The very high similarity in measurement by different phones means any phone can be used.
While some species have distinct wingbeat frequency distributions, the frequency of a few species overlap. The researchers were able to correctly classify the species when the frequencies were fairly unique and did not overlap with frequencies of other mosquito species, while those species with high degrees of overlap were most likely wrongly identified. “This sets the intrinsic limits on the success of any classification scheme based exclusively on wingbeat frequency,” they write.
Besides sound, when they included location parameter datapoint of 20 species from over 200 countries, the accuracy of classification improved not only for individual species but for all species studied. For example, there is 38% probability of misclassifying Anopheles atroparvus with Anopheles dirus. Once the location data is incorporated, it becomes easy to correctly identify the two species — An. atroparvus is found in Europe while An. dirus is found in south Asia.
Similarly, using the time of recording will help in correctly identifying species that show as high as 95% overlap in frequency distributions. Anopheles gambiae and Aedes agypti show 95% overlap but the classification accuracy improves when time of recording is included — Ae. Agypti (which causes dengue, chikungunya and Zika) is active during daytime while An. gambiae (malaria-causing mosquito) is active during the night.
“We hope that a citizen science approach to mosquito surveillance based on this method will boost our capability to dynamically assess mosquito populations, study their connections to human and environmental factors, and develop highly localized strategies for pre-emptive mosquito control,” they write.