BYU researchers create algorithm to predict suicidal thoughts in adolescents

SALT LAKE CITY — If counselors could predict which kids are most likely to commit suicide, they could save lives. Researchers from Brigham Young University, Johns Hopkins and Harvard have come up with an algorithm that predicts suicidal thoughts and behavior among adolescents. Their findings could lead to more effective measures to prevent suicide.

The team of computer science professors and public health professors was able to pretty accurately predict which adolescents will exhibit suicidal thoughts or behavior. They’re doing the research because it’s important to them and the community.

“I’ve had close friends that have had suicides in their family. For me, it’s kind of personal that way,” said BYU computer science professor Quinn Snell. “Our lab had been looking at suicidality and suicide ideation for quite a little while.”

He and his team analyzed state data from nearly 180,000 Utah students who took a student health and risk prevention survey from 2011 to 2017. They applied various algorithms and found a machine-learning model that predicts suicidal thoughts and behavior among adolescents with 91% accuracy.

Just as important, they discovered risk factors that are leading predictors of suicidal thoughts and behavior.

“The number one predictor of that was bullying,” Snell said. “Cyberbullying, and bullying at school and how safe you feel at school. But, right behind that was how safe you feel at home.”

Being threatened or harassed online, or being bullied at school put kids at higher risk for suicidal thoughts or behavior. But, so did exposure to serious arguments and yelling at home.

“This is something that sometimes as parents and school counselors we can’t see,” Snell said.

The researchers believe this is critical knowledge in our state because suicide is the leading cause of death among Utah adolescents, with an average of 660 suicide deaths a year.

“Maybe in communities, maybe in families there are certain things we might want to think a little bit more about. Maybe additional resources or additional energies might need to go to some of those areas,” said Carl Hanson, a BYU professor of public health.

But, not far behind those top three risk factors, they discovered adolescents who regularly eat together with their families had less risk.

“This could be something where you could target a program saying, eat a meal together, spend time together, but don’t spend time arguing together,” Snell said.

They believe the implications for our communities should lead to the development of better programs for suicide prevention. Programs could focus on helping adolescents develop stronger connections with their peers and their families.

“Those are huge,” Hanson said. “So, if communities are going to wrap their head around what to do, that could be a place to start.”

This research could have a big impact in school counseling, but also in clinical counseling as our community steps up to prevent suicide.

“I think our research speaks to the need of additional energy in certain areas of risk,” Hanson said.