Service for acoustic diagnostics of respiratory diseases and COVID-19
The service uses artificial intelligence algorithms to analyze a cough record made using a mobile application or web service. Then it detects the presence of the virus by comparing it with coughs from the dataset. More than 10,000 cough samples were collected together with doctors and employees of City Clinical Hospital No. 40, as well as open datasets. The developers have designed and trained an AI model that recognizes the presence of the COVID-19 virus in the body from an audio recording of a cough.
The model distinguishes the cough of an infected person from a healthy person in 91.5% of cases. The most important fact is that markers of the disease are found in those cases when it proceeds without symptoms. As of March 2021, the service is under testing.