Russian Experts Created Neural Network for COVID-19 Cough Recognition

Russian developers have created an algorithm to diagnose COVID-19 by the sound of a cough. The technology works on the basis of a neural network that has been trained to listen to thousands of recordings of the coughs of those infected with the coronavirus and patients with other diseases, Izvestia reports.  

“The recorded sound can be represented as a picture – a graph of frequency fluctuations. This record is called a spectrogram. Even with the naked eye, you can see a pattern characteristic of COVID-19 on it. This is convenient for doctors who do not want to once again contact a potentially infectious patient,” a researcher at the Lebedev Physical Institute of the Russian Academy of Sciences, Dmitry Mikhailov, said.

It is noted that the device looks like a smartphone. You need to download a special application and start coughing into the speaker. Then the programme will understand whether a person is sick with COVID-19 or not. This technology can be used, for example, to check the cough of citizens in public transport. Currently, the developers are waiting for the approval of the Federal Service for Surveillance in Healthcare.

Ru-Main, 29.10.2020  

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *