AI Can Spot ‘Covid Cough’ That’s Inaudible To Humans
It’s pretty much impossible to cough at the moment without terrifying yourself and anyone around you. But a new AI could help us spot whether or not a cough is actually COVID-related.
The algorithm, which was developed by a team at MIT, can apparently successfully identify people with COVID-19 based solely on the sound of their cough.
In the tests, the algorithm achieved a 98.5% success rate in correctly identifying people who had tested positively for coronavirus, with an even more impressive 100% rate in people with no other symptoms aside from the cough. Researchers say that the critical difference that characterises a ‘Covid cough’ could not be heard by human ears.
Results from the trials were published at the end of September in the IEEE Journal of Engineering in Medicine and Biology, with researchers saying the next step is to get regulatory approval to develop the algorithm into an app.
Brian Subirana, an MIT scientist who co-authored the paper, told BBC News:
The way you produce sound changes when you have Covid, even if you’re asymptomatic.
Practical use cases could be for daily screening of students, workers and public, as schools, jobs, and transport reopen, or for pool testing to quickly alert of outbreaks in groups
MIT is not the only institute working on the use of algorithms to help diagnose coronavirus. A project run by Cambridge University over the summer reported an 80% success rate in identifying coronavirus patients based on breath and cough sounds, while an Israeli startup has developed an algorithm that uses voice samples to spot coronavirus.
This latest algorithm has conducted tests at a much larger scale than other programs, with the MIT lab collecting about 70,000 audio samples, 2,500 of which were from people with confirmed positive coronavirus tests. In the paper explaining their findings, the team claimed that they had collected ‘the largest audio COVID-19 cough balanced dataset reported to date’.
If developed into an app, the AI-based algorithm would likely not be used to diagnose patients by itself, but could be used to help determine whether or not people experiencing symptoms should be sent for further tests.
Calum Chace, an Artificial Intelligence expert, added ,'[It’s] a classic piece of AI. It’s the same principle as feeding a machine a lot of X-rays so it learns to detect cancer, it’s an example of AI being helpful. For once, I don’t see a lot of downside in this.
If you have a story you want to tell, send it to UNILAD via [email protected]
Most Read StoriesMost Read