Skip to main content

The Saskatoon Police are spearheading a new, high-tech way of looking at missing persons cases that the service hopes might one day identify people who are at risk before they disappear.

The Saskatchewan Police Predictive Analytics Lab, which was set up in 2015, is a partnership between the University of Saskatchewan, the provincial government and various public safety agencies. Its goal is to centralize different types of data from police, social services and social media into a hub. It would then analyze the information and deliver it in a quicker, more efficient way during a missing persons investigation.

The program promises to disrupt how police deal with missing people, as law enforcement is increasingly exploring the potential of artificial intelligence as an investigative tool.

While it has yet to be used in the real world, an analysis of the program’s first year, funded by Defence Research and Development Canada, found that the work done thus far has generated interest and excitement from police.

Keira Stockdale, a psychologist who works with the lab and wrote the one-year review, said they are still in the process of testing the database, but are already learning a lot.

The project uses a visual dashboard so officers can view a grid of all recent missing persons cases, with valuable information such as gender, age, and whether they were in the provincial foster system, and how that may link to other open cases.

“While this may seem like basic information, currently those data are not necessarily at your fingertips,” Dr. Stockdale says.

From there, she says, technology like machine learning and artificial intelligence – or even more simple software, like heat maps – can tell investigators a lot.

“The information can be helpful in identifying trends, tracking high-risk cases, and intervening before something happens,” she says.

Trend analysis of unsolved cases has been identified as a large and deadly gap in Canadian policing, going back to notorious serial killers Paul Bernardo and Robert Pickton. More recently, Toronto Police have had to answer for how eight queer men were allowed to go missing over seven years before they realized a serial killer was targeting men of similar sexuality and ethnicity.

The idea of intervening before someone goes missing is particularly novel. Dr. Stockdale says that analyzing the data could show that there is actually some rhyme and reason to why some people disappear.

Dr. Stockdale says the technology could be used to identify the drivers behind why people go missing.

One of the project’s early priorities is to try to prevent disappearances from children in provincial care and so-called “habitual runaways.” Across Canada, but in Saskatchewan in particular where Indigenous youth are vastly over-represented in foster care, there has been a recurring problem of youth running away and dying, through accident, homicide or suicide.

Good data can tell you a lot about those cases, Dr. Stockdale says.

“Sometimes they’re not running from things, they’re running to things – and maybe specifically at certain times of the year, or there’s certain riskier times of day, or times of the week.”

There has long been concern about police forces using artificial intelligence, and the Saskatoon team is aware of that. Dr. Stockdale’s project keeps their databases firewalled from other police systems – meaning that while they receive information from social media and social services, they won’t be sharing that information with the rest of the police force.

Other agencies worldwide are attempting to leverage technology in similar ways. The International Centre for Missing and Exploited Children has recently launched GMCNgine, a program that uses AI and machine learning to scour the internet for pictures of missing children. Mike Cachine, chief technology officer for the centre – which assists law enforcement around the world – says that while using technology to predict and prevent runaways may seem futuristic, “it’s extremely feasible.”

“My goal has been to take what they had, in a static database,” he said, “and apply technology to it, making it smarter, making it faster, making it more proactive – teach this thing how to act like a parent looking for their kid. A parent would never give up, and we don’t want our technology to either.”

The technology that Mr. Cachine works with has also showed some promise in collecting evidence that can lead to the return of missing children – scraping photos taken around the same time, near the same location as where a youth disappeared, for example, which can help identify a suspect’s vehicle or description.

Last year, Amazon awarded missing persons app Second Alert first prize in the artificial intelligence hackathon. The app lets individuals upload photos of missing relatives, and tries to match that photo with entries in global missing persons databases.

“People get lost twice,” Horacio Canales, the app’s creator, said after he won the prize. “First, when they are reported and are being sought, and second, when their disappearance end up piled in a database.”

But those behind the projects agree that predictive technology should only ever augment human police work, not replace it.

“It’s a piece of the puzzle," Mr. Cachine said.