The neural network was taught to determine the presence of coronavirus in the body by coughing – Knife

Researchers at the Massachusetts Institute of Technology (MIT) have developed an algorithm that detects the presence of coronavirus in the body by the sound of a cough, even if a person has no symptoms. About it reported on the university website.

According to scientists, the cough of a patient with asymptomatic covid is slightly different from the cough of other people – the algorithm was taught to find this difference. The program is still being finalized, but soon it will be released as an application for a smartphone – with the help of it, it will be possible to carry out diagnostics even at home.

The program works with the help of neural networks that first assess the strength of a person’s vocal cords, and then compare the new cough record with tens of thousands of cough records of other people with COVID-19 or other pathologies of the respiratory system.

Tests have shown that artificial intelligence detects coronavirus cough with 98.5% accuracy. The developers are ready to implement the neural network as soon as possible, but for now they must obtain permission from the US Food and Drug Administration.

–

–