JS not enabled Artificial intelligence identifies 'Covid cough' inaudible to humans | Daily Manchester
  • Artificial intelligence identifies ‘Covid cough’ inaudible to humans
Artificial intelligence identifies ‘Covid cough’ inaudible to humans

Algorithm achieved a 98.5% success rate among people who received a positive official result on the coronavirus test

 

An algorithm developed in the United States correctly identified people with Covid-19 just by the sound of their coughs.

 

In tests, he achieved a 98.5% success rate among people who received a positive official result on the coronavirus test, increasing to 100% in those who had no other symptoms.

Researchers would need regulatory approval to make it an app.

 

They said the crucial difference in the sound of a cough from an asymptomatic patient from Covid-19 cannot be perceived by human ears.

 

The artificial intelligence algorithm was developed in the Massachusetts Institute of Technology (MIT) laboratory.

 

MIT scientist Brian Subirana, co-author of the study, published in the IEEE Journal of Engineering in Medicine and Biology, said: “The way you make sound changes when you have Covid-19, even if you are asymptomatic.”

 

The mechanism could be used “for daily screening of students, workers and the public, in schools, work and transportation to reopen. Or even for pool tests to quickly alert and prevent outbreaks in groups,” says the report.

Several organizations, including Cambridge University, Carnegie Mellon University and the UK health start-up Novoic, have been working on similar projects.

Sound samples

In July, the Cambridge University Covid-19 Sounds project reported an 80% success rate in identifying positive cases of coronavirus based on a combination of breathing and coughing sounds.

 

In May, he had collected 459 coughing and breathing samples sent by 378 members of the public – she now has a bank of about 30,000 recordings.

 

The MIT laboratory collected about 70,000 audio samples, each containing a series of coughs.

 

Of these, 2,500 are people with confirmed cases of coronavirus.

Leave a Reply

Your email address will not be published.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