MIT’s AI Detects Asymptomatic COVID-19 Coughs Through Sound

“There’s, in fact, sentiment embedded in how you cough,” Brian Subirana, a research scientist at MIT, tells designboom. This knowledge informs a new AI that Subirana’s lab is currently developing for use in COVID-19 detection. Using information gleaned from forced-cough recordings, the AI “distinguishes asymptomatic people from healthy ones.” While the nuances in sound cannot be identified by human ears, the AI can detect asymptomatic carriers with up to 98.5% accuracy. When available—likely through app integration, and only once approved by the FDA—users will simply cough into the microphone on their smartphone. Ideally, the system could be used to screen people before entering public and semi-private spaces. While not a formal test, this could be a valuable, accessible and free weapon to help in decreasing the spread. Read more at designboom.

Image courtesy of Christine Daniloff / MIT