Artificial Intelligence system can predict school violence

Scientists have developed an artificial intelligence system that can predict which students are most at risk of school violence. The researchers found that machine learning – the science of learning computers over time without human intervention – is as accurate as a team of child and adolescent psychiatrists, including a forensic psychiatrist, about the risk of violence at school. “Previous violent behavior, impulsivity, academic problems and negative attitudes correlated with the risk for others,” said Drew Barzman, child psychiatrist at the Cincinnati Children’s Hospital Medical Center in the United States. , UU

“Our risk assessments focus on predicting any kind of physical aggression in the school, and we do not collect results data to assess whether machine learning can really help prevent school violence, it’s our next target,” said Barzman. the study published in the journal Psychiatric Quarterly. The researchers evaluated 103 adolescent students at 74 traditional schools in the United States who had greater or lesser changes in their behavior or aggression toward themselves or others. The students were recruited from outpatient psychiatric clinics, inpatient facilities and emergency services. The team carried out risk assessments with the participants at the schools. The sound recordings of the evaluations were manually transcribed and commented on.

It turned out that after two scales developed and validated by the team in previous surveys, students were equally divided between medium to high risk and low risk. There were significant differences in the overall scores between high-risk groups and low-risk groups. The self-learning algorithm developed by the researchers reached an accuracy of 91.02%, which is considered excellent when interview content is used to predict the risk of school violence. The rate rose to 91.45 percent when demographic and socioeconomic data was added.

“The machine learning algorithm, based solely on the participant’s interview, was almost as accurate in assessing the risk levels as a complete evaluation by our research team, including collecting information from the participants.” Parents and school, a review of the records, though they are available and qualify on both scales, “said Yizhao Ni, a computer scientist from Cincinnati. “Our ultimate goal, if supported by research, is to expand the use of machine learning in schools in the future to increase the structures, professional judgment to more effectively prevent school violence,” said Barzman. , ,