Artificial intelligence has been a useful tool in the fight against the Corona pandemic in recent months. Whether in the development of the new vaccine, patient care or the detection of pathogens. Now MIT has announced another breakthrough in AI. According to a recent announcement, the researchers have succeeded in training an artificial intelligence that can tell whether a person is infected with COVID-19 by coughing.
This is possible even before the affected person shows further symptoms, which makes an early treatment possible. The neuronal networks of the AI recognize the smallest warnings and changes in the person’s cough and thus make the diagnosis.
The network was trained with thousands of cough samples and spoken samples of infected and non-infected persons. The neural networks, the report continues, can now detect the volume of the cough, the emotional state of the person and changes in the respiratory system and, based on the trained patterns, make a diagnosis that was previously 98.5% correct.
As the next step, the MIT research team is now working on integrating artificial intelligence into a user-friendly app to enable free and user-friendly pre-screening. Similar to the Corona app, the user should be able to check daily whether everything is okay or whether it is time for a test.
The real effectiveness of the artificial intelligence must first be demonstrated in the field test, since the laboratory conditions are of course always different than in practice. But the findings are a path in a promising direction.