Researchers turn to AI to flag signs of mental illness online
Computer scientists at U of O developing software to detect 'key words' on Facebook, Instagram, Twitter
Computer scientists at the University of Ottawa are turning to artificial intelligence, or AI, to search for signs of mental illness among users of social media.
The researchers are developing AI software that scours platforms including Facebook, Instagram and Twitter. The software is programmed to flag posts for warning signs of mental health issues such as depression, addiction and suicide.
We are trying to catch early signs [of mental illnesses] and solve them before they become a big problem.- Diana Inkpen, University of Ottawa
Diana Inkpen, a professor of computer science at the University of Ottawa, is heading up the project. She said the software could become a tool for health care professionals to monitor patients through their social media posts.
"We are trying to catch early signs [of mental illnesses] and solve them before they become a big problem. That would be a big gain," Inkpen said.
Key word search
The words and phrases were chosen from a list compiled by medical professionals.
The tool has been put to the test during the Bell Let's Talk mental health awareness campaign, when it was able to scan Twitter and extract tweets containing the key words.
According to Inkpen, each time the software is fed new data it becomes more efficient at gathering specific information and making predictions.
Inkpen is hoping the AI software could be used by public health agencies to gather data on specific areas of mental health, to pinpoint communities that are struggling with mental illness and to help direct funding to where it's most needed.
Interest from doctors
Although the software is still in the research phase, some health care professionals are already interested in using it.
Dr. Gail Beck, the clinical director of the youth program at The Royal Ottawa Mental Health Centre, said she already uses Twitter to communicate with some patients, and believes the AI software could become a beneficial tool.
Inkpen said a $23,000 grant from the Natural Sciences and Engineering Research Council of Canada will go toward funding the last stage of her research, which she expects to finish with the next year.