Researchers are currently testing an app that is supposed to immediately recognize people suffering from Covid-19 – and that by the sound of coughing. Initial tests of the AI app are very promising and the hit rate is already exceptionally high. Everyone can help to make the app better.
Coronavirus: App focuses on cough detection
The Massachusetts Institute of Technology (MIT) has presented its results on an AI-supported app that can very quickly recognize people suffering from Covid-19 by the sound of coughing. The hit rate for people with Covid-19 symptoms is 98.5 percent, the researchers report. The situation is even better in people who do not (yet) show any symptoms themselves, but who already carry the virus. Here, the app was able to successfully determine who is infected and who is not in 100 percent of cases.
<An artificial intelligence can tell by the coughing sound alone whether a disease is present. There would be measurable differences in noise, according to a report published in the IEEE Open Journal of Engineering in Medicine and Biology . These differences are not perceptible to the human ear.
For the required data sets, the researchers had collected 5,000 cough records between April and May and thus were able to build up a database of acoustic biomarkers. Ideally, the app now planned could ensure that more invasive tests do not have to be carried out at the very beginning of a suspected case, which would also take much longer to obtain a result. The app could be used especially in places where many people are present at the same time, such as in companies or educational institutions.
For the coronavirus, the AHA rules must be observed. More about this here in the video:
Coronavirus app: Anyone can help
The MIT uses the “Open Sigma” program to call for action. Volunteers can record their cough on a website set up for this purpose, and then answer a few general questions as well as specific health questions.
Read the original article here.