New AI model detects asymptomatic Covid-19 infections through device-recorded coughs by Massachusetts Institute of Technology (MIT)

For more information read the article: https://news.mit.edu/2020/covid-19-cough-cellphone-detection-1029

TRANSCRIPT:

[AUDIO RECORDING OF A PERSON COUGHING]

NARRATOR: Asymptomatic people who are infected with Covid-19 exhibit, by definition, no discernible physical symptoms of the disease. But it seems those who are asymptomatic may not be entirely free of changes wrought by the virus. The differences between a cough of an asymptomatic patient and a healthy individual are not decipherable to the human ear, but it turns out that they can be picked up by artificial intelligence. 

NARRATOR: For example, here is a cough of a healthy individual:[AUDIO RECORDING OF A HEALTHY INDIVIDUAL]

NARRATOR: And now here is a cough of an asymptomatic person with Covid-19: [AUDIO RECORDING OF AN ASYMPTOMATIC COVID-19 POSITIVE PATIENT]

NARRATOR: To make things even more challenging, listen to a person who has symptoms and is Covid-19 positive: [AUDIO RECORDING OF A COVID-19 POSITIVE PATIENT WITH SYMPTOMS]

NARRATOR: It is very hard, frankly almost impossible, for a person to distinguish these three cough’s, even after you’ve listened to them multiple times. But a team of MIT researchers report they have developed an AI model that can distinguish asymptomatic people with Covid-19 from healthy individuals without the disease through forced-cough recordings. 

NARRATOR: To develop their model, the researchers used tens of thousands of samples of coughs submitted by people voluntarily though web browsers and devices such as cellphones and laptops. When they fed the model new cough recordings, it accurately identified 98.5 percent of coughs from people who were confirmed to have Covid-19, including 100 percent of coughs from the asymptomatic, who reported they did not have symptoms but had tested positive for the virus.

NARRATOR: When the AI model is fed the cough of a Covid asymptomatic person [AUDIO RECORDING OF AN ASYMPTOMATIC COVID-19 PATIENT] they found it was able to pick up patterns in the four biomarkers — vocal cord strength, sentiment, lung and respiratory performance, and muscular degradation — that are specific to Covid-19.

NARRATOR: When the model is fed the cough of a covid-positive individual who IS exhibiting symptoms [AUDIO OF A SYMPTOMATIC COVID-19 POSITIVE PATIENT] it is actually harder for artificial intelligence to discriminate. The researchers think it is because there are many conditions that create symptoms, such as the flu or asthma and therefore the results are confounded. For this reason, they stress that their AI model is not meant to diagnose symptomatic people, OR determine whether their symptoms are due to Covid-19 or other conditions. The tool’s strength lies in its ability to discern asymptomatic coughs from healthy ones.

NARRATOR: The team is now working on incorporating the model into a user-friendly app which if FDA-approved and adopted on a large scale could potentially be a free, convenient, noninvasive prescreening tool to identify people who are likely to be asymptomatic for Covid-19. A user could log in daily, cough into their phone and instantly get information on whether they might be infected and therefore should confirm with a formal test.