A new algorithm could detect ‘Covid cough’ - even in people with no obvious symptoms

An algorithm has been developed which is apparently able to correctly identify people with Covid-19 by the sound of their coughing.

Sign up to our daily newsletter

The i newsletter cut through the noise

People suffering with the virus, even if they are asymptomatic, produce sound differently to those who aren’t infected, according to researchers. But the difference between the coughing of someone with Covid, and without, can’t be heard by the human ear.

Tests have found that the algorithm has a 98.5 per cent accuracy rate with people who had already received a positive diagnosis, but was 100 per cent accurate for people with no other symptoms.

How does the algorithm work?

The algorithm works by analysing around 70,000 audio samples of people with and without Covid breathing and coughing, and looking for patterns, then applying that data to new audio samples.

If it can be rolled out successfully, the algorithm could have a major impact on how societies can live with the virus.

According to a report in the IEEE Journal of Engineering in Medicine and Biology, “Practical use cases could be for daily screening of students, workers and public, as schools, jobs and transport reopen, or for pool testing to quickly alert of outbreaks in groups.”

The technology was developed at the Massachusetts Institute of Technology (MIT) lab using artificial intelligence to create the algorithm, which will be able to be developed into an app once it is regulated.

This project used some of the research from a separate algorithm which analysed voice recordings to try and identify Alzheimer’s disease, which weakens the vocal cords.

Other organisations have been working on their own versions of this kind of technology, including Cambridge university, but MIT’s is much more advanced and accurate at this stage.