Researchers at the Massachusetts Institute of Technology have developed artificial intelligence that can identify cases of COVID-19 based on audio recordings of coughs. The AI model it was able to identify 98.5% of coughs from people who had confirmed cases of the virus, including 100% of the coughs from those reportedly did not have symptoms but still tested positive.
Researchers at the Massachusetts Institute of Technology believe they've found a way to determine if people have coronavirus based on the sound of their cough through artificial intelligence.
The human ear is incapable of picking up on the acoustic variations of a coronavirus cough, but AI can, the researchers say.
Get top local stories in Boston delivered to you every morning. Sign up for NBC Boston's News Headlines newsletter.
The technology could be a could indicator of whether or not someone should get tested for the virus, even if they don't have symptoms.
A team of researchers began collecting audio recordings of coughs in April. The samples were voluntarily submitted on a website using a cell phone and laptop.
Through an AI model, the technology was able to identify 98.5% of coughs from people who had confirmed cases of the virus, including 100% of the coughs from those who reportedly did not have symptoms but still tested positive.
Developers are now building an app to allow people to use something as convenient as their smartphones to access the technology and find out if they could be infected. The model still needs federal approval.
The group of researchers were working on a similar model for Alzheimer’s before the pandemic, according to MIT, when they decided to see if it could also work for COVID-19.
