The algorithm that can detect the ‘Covid cough’

Today in “isn’t technology amazing?!” news, we thought we’d talk about something a little different. Even when it seems like the world is on fire, there’s always a silver lining. The silver lining we want to focus on right now is how the tech industry is developing myriad, innovative ways to tackle the coronavirus. Although these developments are focusing on coronavirus right now, they could very well be used to detect other illnesses in the future. The first is the Covid cough algorithm.

Diagnosis via an app?

Researchers at the Massachusetts Institute of Technology (MIT) have developed an Artificial intelligence model that can detect people who have Covid-19 simply by the sound of their coughs. It’s believed that the new algorithm will be especially helpful for spotting the virus in asymptomatic people who have no other symptoms. 

Even without the tell-tale symptoms, MIT researchers have found that asymptomatic individuals with coronavirus cough differently to healthy individuals — a difference in sound that cannot be picked up by the human ear.

The team made the discovery through a study in which the participants submitted forced-cough recordings via a web browser on their cell phones or laptops. The AI model was then trained through thousands of sample coughs and spoken words. The AI recognized 98.5% of people who were confirmed to have Covid-19. It also identified 100% of asymptomatic people who had tested positive for the virus but had no symptoms. 

The ultimate goal is to create an app using the new AI model. Such an app could be used as a prescreening tool for those who have potentially been exposed to the virus before getting a test. 

Your Fitbit may also help detect Covid

Meanwhile, researchers at Stanford University have been studying the possibility of wearables that track heart rate (such as Fitbits) detecting Covid-19 before symptoms even appear. Although it may take several days for someone exposed to the virus to show symptoms, there is a presymptomatic period where the body may give off other signals, such as a sudden change in resting heart rate. The researchers produced an algorithm that detected heart rate abnormalities in 56% of Covid patients tested in the days before their symptoms started. 

Wrap Up

While it’s early days for both of these research projects, it’s heartening to see such interesting health tech developments during a time relatively void of good news. It’s also likely they could be repurposed to diagnose other illnesses in the future, especially when you consider that the MIT study was based on an initial AI framework for detecting Alzheimers.

Share on Twitter, Facebook, Google+