Predicting the Next Pandemic by Counting Coughs

Sep 4, 2020

 Many researchers at Virginia Tech are switching gears to focus on fighting coronavirus. 

One project aims to provide early warning that another pandemic might be on the way by analyzing people’s coughs.

 

Virginia Tech researchers are working on a scalable, highly deploy-able, cost-effective approach for monitoring respiratory illnesses that could directly be used for public health epidemic forecasting.
Credit Peter Means for Virginia Tech

Not all coughs are the same.   Used to be, you heard someone cough and you think little of it. But today, you could clear a room with a cough.

 

Shane Ross teaches engineering at Virginia Tech. “We're trying to develop a system that will automatically detect and count coughs so that we can generate a coughing signal over time in public places. The findings will be correlated with influenza like illness or other respiratory disease indicators.”

 

You might ask, what’s a professor of aerospace and ocean engineering doing counting coughs? Well, the coughs are the data.

 

“And it can be used as a signal for predicting (rates of) respiratory illness across the country. So, we're first developing it in one place here at Virginia Tech, probably in classrooms. And then we'd like to deploy it elsewhere.”

Ross and colleague, Nathan Alexander, an expert in machine learning, will create a ‘Library of Coughs’ by putting recording devices in public places.

 

“And this would be part of the later comparison of not only is it a cough?  But also, but what specific illness is it connected to?

 

Ross says counting and analyzing coughs amounts to what’s known as an “independent signal” of disease – a marker completely separate from the data you get from direct virus testing and ideally, ahead of it, for early warnings. So, the thought is that, yes, this could, if people have, are coughing, say a few days before they actually go into the doctor, it provides a little bit of lead time. But I think the more important thing is its a, it's an independent signal.”

 

The idea is to gather data but protect people’s privacy

 

“This is the sensitive stuff” says Ross. "We recently applied to the local ethics board. Our goal is to do this in an unintrusive way that protects people's privacy. We're not trying to single out sick people. We want to listen in areas where there are large numbers of people.”

 

The project is in the early phases.   Right now, they’re working to develop the automatic sound detection software. If it’s successful, it could become part of a national infectious disease prediction center that may be in the works. There is growing interest in creating a center.

 

Virginia Tech’s COVID-19 Rapid Response Seed Fund gave the team $50,000 to get started, and there is a crowdfunding campaign to support projects like these going forward.

 

Ross says one of the signals they have now are these influenza illness levels. “That’s measured at the state level, but it would be nice if there was something much more immediate and in real time, that you could get anywhere.”

***Editor's Note: Radio IQ is a service of Virginia Tech.