MIT Develops AI Tool To Detect COVID By Listening To Coughs 

MIT Develops AI Tool To Detect COVID By Listening To Coughs 

Tyler Durden

Fri, 11/06/2020 – 22:40

MIT Researchers have developed an AI pre-screening tool that can accurately detect if someone has COVID-19 via audio of their cough. The breakthrough technology could be embedded into a smartphone app and distributed to the general population, used even before clinical testing to recognize if someone is infected. With the second coronavirus wave arriving in Europe and the US, the AI pre-screening tool could be an effective bet to screen for asymptomatic patients.

The paper, titled “COVID-19 Artificial Intelligence Diagnosis using only Cough Recordings,” was recently published in the IEEE Journal of Engineering in Medicine and Biology, specifies that the AI pre-screening tool distinguishes asymptomatic people from healthy individuals through forced-cough recordings. 

Lead researcher of the study Jordi Laguarta, and co-researchers Ferran Hueto and Brian Subirana, along with their team, used the AI tool to test tens of thousands of audio cough samples. Their AI tool had a 98.5% accuracy rating for identifying coughs from people with COVID-19, including 100% of coughs from asymptomatic people. 

Visual Summary: The AI model architecture shown enables a non-invasive, real-time solution for an unlimited throughput daily country-wide screening, outbreak monitoring, and viral/serology test pooling candidate selection at essentially no cost. As shown on the ROC curve, the model discriminates officially tested COVID-19 subjects 97.1% accurately with 98.5% sensitivity and 94.2% specificity, with a 100% asymptomatic detection rate and 88% accuracy on all subjects. These results are based on a dataset of 5320 subjects. h/t COVID-19 Artificial Intelligence Diagnosis using only Cough Recordings 

According to MIT News, the researchers are planning to integrate the new AI tool into a smartphone app that would allow users to log in on a daily basis to record a cough and get real-time results of whether they’re infected or not. Upon FDA-approval, the app could be a transformative pre-screening tool. 

The researchers said, “the effective implementation of this group diagnostic tool could diminish the spread of the pandemic if everyone uses it before going to a classroom, a factory, or a restaurant.” 

They concluded: “Pandemics could be a thing of the past if pre-screening tools are always on in the background and constantly improved.” 

What’s scarier than contracting the virus is the encroaching surveillance state. If the app is packaged up, cleared by the FDA, and distributed the millions, who exactly will be receiving this data?

Will an app user who is flagged for a possible infection be monitored via the smartphone’s GPS?

These are some of the questions readers should be asking. 

Go to Source
Author: Tyler Durden

Comments