A technology originally conceived to remotely monitor the heart rates of premature babies has been transformed into a potential way to scan huge groups for illnesses like COVID-19.
A Canadian company is adapting the technology for drones and closed-circuit video cameras.
But being able to scan people from up to 50 metres away and tell if they’re sick, potentially without them knowing, raises some big privacy concerns.
The technology connects video cameras, including those equipped with thermal-imaging systems, to a computer program with learning algorithms.
The system can monitor temperature, heart and respiratory rate, and tell when someone is sneezing or coughing.
The computer can determine if someone has a respiratory illness or infectious disease, according to Javeen Chahl, a professor of engineering at the University of South Australia.
Chahl headed up the research developing the system.
The more video the program has of people sick with a particular disease, the better it is at detecting signs of that illness.
To get more footage of people with COVID-19, a pilot project using the system is expected to be run in the New York City area in the coming weeks.
The technology is attracting attention in Canada, where some airports have expressed an interest, according to Canadian drone manufacturer Draganfly Inc.
Chahl has partnered with Draganfly to expand the use of his research. The company wants to sell the technology to governments, along with medical and commercial customers.
Draganfly has a manufacturing plant in Saskatoon and offices in Vancouver and the U.S. Much of its business has come from selling drones to first responders.
Both Chahl and Draganfly hope this technology will provide a quick way to screen for COVID-19 and help detect the spread of other diseases.
“This technology can be used … even as a global early-warning system,” said Cameron Chell, the CEO of Draganfly Inc.
“It’s really designed to get stats on what’s actually happening, so the real decisions can be made and policy can be set.”
Chell said the cost of the pandemic has already been too high with thousands of deaths worldwide and billions of dollars in economic loss.
He said being able to get timely information on how many people are sick and where they are located are ways to help governments and health authorities stop the spread of infectious diseases.
But developing this kind of technology, especially during an emergency, is a concern for David Fraser, a privacy lawyer with McInnes Cooper in Halifax.
Fraser said sometimes companies and governments rush to develop technologies or procedures during a crisis without full consideration of public privacy or security.
He said proper checks and balances need to be put in place before any kind of new scanning technology is used.
“To widely deploy a technology to collect baseline health information about a population, and about its individuals who are part of that population, that needs to be justified. It needs to be justifiable and that needs to be a societal conversation,” said Fraser.
“It has generally been my experience that the people who make the tools, or deploy the tools, or sell the tools are not the ones in a good position to evaluate the privacy impact of those tools.”
Chell understands those privacy concerns.
He said the system is not designed to single out individuals, but to monitor the health of a population. However, that capability might not be far away.
Chell likens the rise of this kind of medical scanning technology to the widespread use of metal detectors and body scanners at airports after 9/11.
“Health protection is as important as firearms protection,” said Chell. “I think this is part of the new reality and we’re going to have to learn how to be OK with it.”
Once the computer scans enough people with COVID-19, it could be used to screen individuals before they come into a hospital or health-care centre, said Chahl.
The system is not meant to replace traditional diagnostic tests, but can provide an additional screening method.
The computer could identify people with symptoms so they can be isolated as soon as they enter a facility.
“If you reduce … essential services personnel to half as much risk then they might be able to keep operating for twice as long before they contract the virus, for example. So it’s a great benefit to filter and screen,” said Chahl.
Equipping a drone with the system would also be helpful, according to Chahl.
He said if an organization was involved in a relief effort in an area hard hit by an infectious disease, a drone could get an idea of the health of the population without endangering anyone.
But that’s not yet possible.
No one is sure how much information would have to be fed into the system before it could correctly identify one illness from another.
It is possible to use computer programs to tell when a person is sick. However, it’s tricky trying to teach a machine how to identify a specific illness, according to Thomas Trappenberg, a professor of computer science at Dalhousie University.
Trappenberg studies machine learning, the mechanism Chahl’s computer program uses to identify sickness.
“They need a lot of data,” said Trappenberg. “How many data we have to feed them is also very hard to predict.
“I think this is what the company is trying to do right now, trying to get more images, and it is a little bit stabbing in the dark. But it’s well worth probably trying it.”
There are other problems that will have to be solved before the technology can be effective, according to Sageev Oore, a computer science professor and machine learning researcher, also at Dalhousie.
“On real streets with real people in real weather conditions, with real movement and real camera lenses that might get a little piece of dirt on them, all of those things add, cumulatively, a lot of challenge,” said Oore.
Even so, Oore thinks the research is worth doing.
Chahl said more needs to be done before the system reaches its full potential.
Still the technology has been used to detect whether people were alive under debris, by comparing the way debris shifted in relation to a person’s breathing frequency. It also succeeded at measuring the heart rate of premature babies in incubators.
Trying to figure out the heart rates of babies without the use of electrodes or other equipment is how the project started back in 2015.
Fraser said even with those successes, it’s important the determinations the computer makes about someone’s health are confirmed by medical testing.
He applauds people who want to use their skills to try to help society cope with this pandemic, but he said people still need to be cautious.