New MIT algorithm can tell if you have Covid-19 from your...

Will the sound of coughing soon be a new indicator of Covid-19? MIT has just answered this question in the affirmative by developing an algorithm to identify people infected with Covid-19 solely on the basis of their cough. This algorithm, which could constitute a turning point in the fight against the coronavirus pandemic, was formed from “tens of thousands” of recordings – coughs and words – and was able to correctly identify 98.5% of people who presented with symptoms and had confirmed cases of Covid-19.

Even stronger, the algorithm developed by the prestigious American university identified 100% of asymptomatic carriers of Covid-19, whose presence of the virus had been confirmed but who did not present any symptoms. The recordings used to train this artificial intelligence (AI) model were submitted by online volunteers and included forced coughs from healthy volunteers as well as people with Covid-19. More than 70,000 samples have been collected so far and around 2,500 have been submitted by people who have been confirmed to have Covid-19.

“Asymptomatic people may differ from healthy people in the way they cough,” commented the team of researchers behind the project. “These differences cannot be read to the human ear. But it turns out that they can be detected by artificial intelligence ”, explain the latter.

An upcoming mobile application?

Building on these positive results, the MIT team is now working on the development of a mobile application to integrate this model and make it accessible to the general public. However, such an application will require regulatory approval from health authorities and cannot be considered a formal diagnostic tool, the researchers warn. The app could nevertheless act as a “non-invasive screening tool” for users before they confirm their suspicions with an official test.

As tracking and tracing methods multiply with varying degrees of success around the world, this tool intended to tackle the problem of asymptomatic transmission could be invaluable in the fight against the spread of Covid-19. “The effective implementation of this group diagnostic tool could decrease the spread of the pandemic if everyone uses it before going to a classroom, a factory or a restaurant”, notes Brian Subirana, the one of the researchers behind the MIT algorithm.

Across the Atlantic, MIT is currently working with a number of hospitals that will provide additional records to further improve the model. He recalls that his AI model is based on previous studies that examined how coughing and changes in speech patterns could indicate other diseases, including Alzheimer’s disease. The research article detailing this new algorithm was published in the academic journal IEEE Journal of Engineering in Medicine and Biology.

Source : ZDNet.com