Since the virus spread worldwide in the spring, teams all over the globe have been working on developing artificial intelligence programs that can successfully diagnose COVID-19 in asymptomatic individuals based on the sound of their cough. This works because, while someone may not have any symptoms that are detectible by human observation, the virus will still cause subtle changes in their body that technology is able to pick up.
“The human ear is capable of distinguishing maybe five to ten different features of cough. With signal processing and machine learning, we can extract up to 300 different distinct features,” said Ali Imran, a leader on a research project at the University of Oklahoma’s AI4Neworks Research Center that developed one of these AI programs.
Other teams working on such AI systems include the COVID-19 Sounds project at the University of Cambridge; Cough Against Covid, funded by the Bill and Melinda Gates Foundation, at Wadhwani Institute for Artificial Intelligence in Mumbai; and the Coughvid project at the Embedded Systems Laboratory of the École Polytechnique Fédérale de Lausanne in Switzerland. These programs, with FDA approval, could be developed as smartphone apps that anyone could use, breaking down significant barriers to widespread COVID-19 testing.
