Artificial intelligence detects corona infections – through coughing!



[ad_1]

Updated: 04.11.2020 – 18:14

New MIT study
Artificial intelligence detects corona infections – through coughing!


Do you cough once in your tablet and do you already know if you have Corona?  A new MIT study gives hope that this may soon become a reality.  Artificial intelligence is supposed to recognize corona infections.

Photo: iStock.com/Courtney Hale

Do you cough once in your tablet and do you already know if you have Corona? A new MIT study gives hope that this may soon become a reality. Artificial intelligence should recognize corona infections.

The whole world is waiting for a vaccination against the new corona virus. But does the true miracle cure for the pandemic come from a completely different direction?


Massachusetts Institute of Technology (MIT) is one of the leading universities in the world. So it probably shouldn’t come as a surprise that a promising new attempt to contain the corona pandemic has its origin here, of all places. What’s even more surprising, however, is the method itself: according to a new study published in the IEEE Journal of Engineering in Medicine and Biology, one Artificial Intelligence Recognize people with corona infection by coughing! It looks promising. But the true strength of the technology promises a big step forward: Even infected people without symptoms should be able to be identified without error.


Artificial intelligence also recognizes infected people who are asymptomatic from cough


According to the research team led by MIT scientists Brian Subirana, Jordi Laguarta and Ferran Hueto, artificial intelligence recognizes 98.5% of those suffering from Covid-19 with symptoms and a staggering 100% of those infected who did not have symptoms. The reason for this, according to scientists, it is believed that any infection with the virus causes changes in the lungs, which in turn affects the sound of the cough. Although the difference is imperceptible to humans, the computer can tell.


The success rates of the study, which MIT has already reported, MIT News, are promising: in particular, early identification of asymptomatic infected people could presumably prevent many new infections. Because these have no symptoms, they usually have no reason to be tested, they don’t find out their Covid-19 status and then pass the virus on to other people.

How can the result be put into practice?


The study looked at more than 70,000 audio recordings on which people coughed multiple times, meaning the sound of 200,000 “coughing” was examined in total. About 2,500 recordings were from people who tested positive for coronavirus, with no symptoms and with symptoms. These recordings were recorded and transmitted using a web browser or smartphone.

The good news: the cough doesn’t have to be genuine. Forced and artificial cough is sufficient for detection. This means that artificial intelligence could theoretically also be used in everyday life: at the moment, researchers are working to further develop the program into an easy-to-use app. If it were to be approved by the Food and Drug Administration (FDA, German: US Food and Drug Administration) and used on a large scale, you could easily test for free every day if you are infected with the coronavirus, even if you have no symptoms.


“Effective use of this global diagnostic tool could reduce the spread of the pandemic if everyone used it before entering a classroom, factory or restaurant,” study author Brian Subirana told MIT News.



The full study can be found here: COVID-19 AI Diagnosis Using Cough Recordings Only


Can artificial intelligence contain the pandemic? We are curious and will keep you updated on our coronavirus topic page. Here we also answer the question of whether the AHA rules are also effective against the flu and report the expert warning that the intensive care units will be full in two to three weeks.


The latest videos from BILD der FRAU

The latest videos from BILD der FRAU

show description


.

[ad_2]
Source link