Scientists from the Massachusetts Institute of Technology (MIT) build an AI model that can diagnose asymptomatic COVID-19 cases using handset-recorded coughs.
Experts in the US discovered that asymptomatic people might vary from salubrious people in the way they cough. A new artificial intelligence (AI) model can recognize and differentiate asymptomatic COVID-19 patients from healthy individuals. The results can be obtained through their cough records, which can be accessed and viewed in an application.
However, the quality of being different is not decipherable to the human ear. But it can be easily taken up by artificial intelligence. They added a statement.
They said volunteers proffered these records through web browsers and other devices like smartphones and laptops.
Noninvasive pre-screening tool to discover COVID-19 patient
The scientist trained the designed new AI model on tens of thousands of units of coughs and spoken words. If they served the cough recordings, it perfectly-identified 98.5% of coughs from individuals who were proved positive to own COVID-19. They did not have symptoms but had tested positive for the virus, according to the researchers.
The team is operating on consolidating the prototype into a user-friendly app, which could be available freely, comfortable, and one of the pre-screening tools to classify people who are likely to have asymptomatic COVID-19, they proclaimed. As per their statement, users could sign in to the app daily, cough into their smartphone, and spontaneously get data on health.
The adequate implementation of such assembly diagnostic tools could lessen the extent of the COVID-19 if it comes into practice so that everyone before going to a classroom, a company, or a restaurant, said Brian Subirana, a research scientist at MIT’s Auto-ID Laboratory.
Researchers have found that asymptomatic Covid-19 patients may have distinctive coughs. Indecipherable to the human ear, the differences can be picked up by artificial intelligence. Cellphone-recorded coughs could provide a convenient screening tool. https://t.co/OyHqC2KqNk pic.twitter.com/0kYdEIXQkd
— Massachusetts Institute of Technology (MIT) (@MIT) October 29, 2020
Earlier, research groups had already been tutoring algorithms on cellphone records of coughs. This is to diagnose conditions precisely, such as pneumonia and asthma.
Likewise, the MIT group was advancing AI models to examine the cough recordings to see if they could detect other signs of Alzheimer’s. The team set out in April to gather more recordings of coughs, including COVID-19 subjects.
COVID-19 survey for finding new subjects
Members can fill out a review of signs and traits they encounter with the COVID-19 test taken in the hospital. They also did note their gender, language, and information related to the nativity.
Till now, they have gathered over 70,000 recordings, said Subirana. It was the latest and largest dataset to gather and process in the laboratory. The group almost used the 2,500 COVID patient’s records, along with other datasets. They worked with 4,000 of these samples to instruct the AI model.
The experts could separate patterns in the four high markers that are particular to COVID-19 — vocal cord strength, sentiment, lung and respiratory performance, and muscular degradation — that are specific to COVID-19. The model recognized 98.5% of coughs from COVID-19 subjects, they said.
Image courtesy of Michael Traitov/Shutterstock