MIT develops system that identifies symptoms of coronavirus via mobile phone

A team of researchers at MIT is developing software capable of identifying the cough caused by COVID-19, even when the patient is asymptomatic.

[dropcap]W[/dropcap]ith the help of artificial intelligence software and neural networks, experts have already managed to develop a model that can distinguish the small differences between the cough of an infected person and that of an uninfected person.

One of the neural networks is allocated for the identification of the patient’s vocal strength, while another one is able to identify emotional states normally associated with a neurological decline, such as frustration or apathy. There is also a third network that assesses the patient’s respiratory performance as he coughs. Finally, an algorithm assesses muscle breakdown and thus completes the set of tools needed for diagnosis.

To date, the AI ​​that articulates all these elements has been highly accurate in its tests, achieving 98% of cough analyzes in patients with a confirmed medical diagnosis of COVID-19, while in the case of asymptomatic patients the system has 100% of the evaluations.