AI diagnoses Covid-19 by cough sound

AI diagnoses Covid-19 by cough sound 3

AI diagnoses Covid-19 by cough sound

After a month of launching the `Covid-19 identification through cough` project, the AICovidVN research team said that the AI ​​algorithm trained by domestic engineers was able to accurately identify 91% of positive cases.

According to research published in January 2021 by experts from MIT University, when new viruses invade the body, they have not penetrated deeply enough to create symptoms such as fever or cough but have caused minor damage.

In Vietnam, AICovidVN’s project is using available data resources with 1,700 cough recordings of positive people (from Switzerland and India) and a number of other open sources, but there is still a lot of noise and noise.

After one day of launching, AICovidVN has collected nearly 100 cough samples of Vietnamese people.

Despite achieving 91% accuracy, for AI to better diagnose patients suspected of being infected with Covid-19, the project needs at least 10,000 cough samples of Vietnamese people.

When new test results are available, the recorder can update the health status through the project reception portal, and engineers will continue to process it to train the AI.

AICovidVN is a non-profit project with the participation of more than 1,100 engineers, technology experts, doctors and volunteers in many fields.

Due to the complicated situation of the epidemic, Vietnam has soon applied AI to diagnose and search for suspected Covid-19 cases in the community.

Another group of Bkav engineers is also working with the Central Tropical Hospital to take samples to train AI to test Covid-19 through physiological saline.

Post Comment