U of R researchers developing automated pain detection system
It sounds like science fiction, but University of Regina researchers are developing cameras to flag when non-verbal adults with dementia are in pain and a notification system to alert staff of their anguish.
For many years, Thomas Hadjistavropoulos, co-leader of the research project, has studied the problem of people with severe dementia, such as Alzheimer’s disease, who can’t communicate that they are in pain.
“You can have somebody with a tooth abscess, which is tremendously painful, and the pain can go undetected for days, sometimes longer,” he said.
People with severe dementia who have untreated pain can turn aggressive. The aggression is attributed to a psychiatric problem and psychotropic medications are prescribed, which don’t treat the pain, but increase the risk of death by falls or stroke.
“For frail people with dementia, psychotropic medications should be used with a great deal of caution,” Hadjistavropoulos said.
He and his graduate students have developed methods for detecting, evaluating and measuring pain in adults with dementia through pain behaviours — specific facial responses that are relatively uncommon in situations other than pain.
A lowered eyebrow or wince are things staff would notice if they had time to watch residents closely, but often resources are limited.
“That got us thinking, what if we were able to detect and monitor these behaviours in an automated way and reduce the need for a nurse or another clinician to go around and look for pain behaviours on a regular basis — what if a system could do that?” Hadjistavropoulos said.
They’ve come up with a system of strategically placed cameras that send information to a computer which analyzes it based on an algorithm.
When the system detects patients who are in pain, nursing staff will be sent an email, and a light at the nursing station would flash indicating an email was sent.
The researchers are just completing year three of the five-year project.
“This sounds like science fiction to a lot of people, but it’s a lot more feasible than people think,” Hadjistavropoulos said.
Algorithms exist to detect pain behaviours, but currently only work for younger people under ideal lighting conditions.
“They will not work for a wrinkled face because it will interpret a wrinkle as a frown,” Hadjistavropoulos said.
The existing algorithm is being modified so it can distinguish between wrinkles and frowns and be used from multiple angles.
“We’re doing it so that it works under less-than-ideal lighting and camera conditions to make it affordable,” Hadjistavropoulos said.
He noted there are no privacy concerns because the system won’t record the video.
“It will just be processing the images to see if there is pain behaviour,” Hadjistavropoulos said.
With patients’ permission, researchers have video-recorded 102 people who showed pain symptoms.
“This resulted in 1,050,000 frames of video — each frame had to be coded by hand for pain behaviours so we can tell the computer system what they are,” Hadjistavropoulos said.
“It was a tremendous amount of work. I had three coders who worked on this over the course of a year.”
The U of R team is sharing its raw data with a collaborating team at the Toronto Rehabilitation Institute, which is writing the code based on the information.
“We expect to have concrete results on the accuracy of this system by July,” Hadjistavropoulos said.
He hopes the system will be built as a prototype in less than a year and then set up in a couple of longterm care homes to trial.
“It’s a neat project and we’re very optimistic this will make a difference,” Hadjistavropoulos said.
The research is funded by the Age-Well Network of Centres of Excellence — a federally funded program that involves 37 university research centres across Canada and the Canadian Institutes for Health Research.
To mark Brain Awareness Week, Hadjistavropoulos will describe the team’s work at a free public lecture being held Tuesday at 6 p.m. at the Research and Innovation Centre on the U of R campus.
He will be joined by two other presenters.
Gretta Lynn Ell, executive director for Continuing Care, Programming and Utilization in the Long-Term Care program with the Saskatchewan Health Authority, will speak about the challenges in nursing homes because of patient pain and patient falls.
Stephen Robinovitch, a professor from Vancouver’s Simon Fraser University, will talk about using technology to prevent falls in older adults.