TORONTO – Chris Poulin knows just how difficult it can be to cope with losing a loved one to suicide.
Coupled with the pain of losing close friends to suicide and years of work in artificial intelligence (AI) research, specializing in linguistics, Poulin and his fellow investigator Paul Thompson began working on a project aimed at seeking real-time predictive analytics for suicide risk.
Dubbed “The Durkheim Project,” the initiative uses AI algorithms to track common words and phrases mentioned on people’s social networks to identify those showing signs of suicide.
After the person opts in to the project, apps – available on computers, iOS and Android devices – monitor what the user posts on social sites such as Facebook, Twitter and LinkedIn and stores the data in a secure medical database.
Once in the database the algorithm begins searching for keywords that would alert doctors if a patient were showing signs of suicidal thoughts.
“We feel like we have proven that we can predict suicide risk based on language and text factors – and the suicide research community has no ability to predict suicide currently,” Poulin, director and principal investigator of The Durkheim Project, told Global News.
“We are intentionally trying to push the social issue; we are trying to be controversial – in a good way – in having people think about these issues. The ultimate goal is helping other groups and helping the medical community get better at predictive suicide and actually reducing it.”
The first phase of research for The Durkheim Project saw researchers build statistical models of the risk of suicide of U.S. veterans – the group at the centre of the initiative.
The issue is also prevalent among Canadian soldiers and veterans.
According to a study released last week, about 13.5 per cent of Canadian Forces soldiers who served in the Afghan mission have mental health issues linked to their deployment.
Common ailments included PTSD and depression – which could lead to an increased risk of suicide.
Phase 2 of the project involved building a data collection framework based on social networks and mobile data, acquiring the help of social networking giant Facebook.
The final phase, in the works now, involves finding a way to facilitate intervention.
“The medical community has no consensus on what suicide prevention will look like – I just know that we have our own designs and our research plans for what we would call automated intervention,” explained Poulin.
“We are trying to facilitate these kinds of intervention phases in the future.”’
The program currently does not offer any sort of intervention to those who do show risk of self-harm.
The team now has a proof-of-concept clinical application in place that doctors can use to view the risk profiles of veterans as they come in.
But the project is still evolving and has faced some criticism due to what some say is a limited study.
The program, as is, is an opt-in-only program – veterans and active duty soldiers who wish to participate must provide consent for The Durkheim Project applications to actively monitor their social media accounts.
Poulin is the first to recognize that opt-in-only programs limit the scope of the study – especially since the topic of suicide still comes with a social taboo – however, he is hopeful that the program could work as a preventative measure.
“I fully recognize that a limitation in the study is that many of the people that are at the highest risk will not opt-in and I know that – but I guess we are taking a stand on the side of opt-in ideologically,” said Poulin, who describes the app as a clinical-screening tool – much like when doctors screen for cholesterol and recommend dietary changes to their patients who show premature risk.
“I do think that there is one clinical “gotcha” that might make it effective, which is that someone might feel healthy a few years before and they sign up then, but as they gradually get worse but now the system can help them because of prior consent.”
Though the program is currently focused on war vets, Poulin notes that there is nothing barring others from opting in to the program.
“It’s just that the medical protocol is focused towards veterans and those in active duty,” he explained. However, Poulin hopes that in developing intervention measures the team can work to expand the program to different groups of people.
“One of the biggest groups that talk to me about this is those concerned about teenage suicide – mothers and family groups that realize that this is a big problem in the teenage group. You really want to help these kids, their parents and their families,” said Poulin.
According to the Canadian Mental Health Association, suicide accounts for 24 per cent of all deaths among 15- to 24-year-olds and 16 per cent among 25- to 44-year-olds.
Comments