Researchers from MIT have developed artificial intelligence who can recognize the sound of coughing from an infected person COVID-19, even though they are asymptomatic.
This is done by developing multiple neural networks that can distinguish the subtle changes that indicate the effects of Covid-19.
One neural network detects sounds related to vocal power. Others listen to emotional states that reflect neurological decline, such as increased frustration.
The third network measures changes in breathing performance. The algorithm then checks for muscle degradation (weaker coughs) and it provides a more complete picture of a person’s health.
The artificial intelligence was very accurate in initial testing. After the research team trained its model on tens of thousands of cough and dialogue samples, the technology recognized 98.5 percent of coughs from people with confirmed cases of COVID-19. This technology also identifies 100 percent of people who appear to be asymptomatic (asymptomatic).
However, there are clear limitations to this invention. This technology is not intended to diagnose people who are symptomatic, as they may have other conditions that produce similar behavior. And while quite capable, these findings cannot be used to determine with certainty whether a person is infected or not.
Scientists are developing a “user friendly” application that can be used as a virus filtering tool. The user may only need to cough into the phone every day to determine if it is safe to go outside.
Researchers have even suggested that the discovery could end a pandemic if the device was always listening in the background, although the proposal could raise privacy concerns.
–
