Study: Can AI identify impending suicide attempts?

Jessica Kim Cohen - Print  | 

A study in BMC Public Health investigated whether identifying behaviors associated with impending suicide attempts could inform future intervention initiatives.

The researchers, led by Brian L. Mishara of University of Quebec at Montreal, conducted two research projects focusing on suicide attempts in public transit systems. The first asked trained observers to watch video surveillance recordings of suicide attempts in underground stations to identify behaviors that indicate suicide risk, such as leaving an object unattended on the platform.

The second research project served to verify whether coding these behaviors could determine whether an individual was at risk for a suicide attempt. The researchers mixed 63 video surveillance recordings of suicide attempters with 56 video surveillance recordings of other passengers, to determine if students using the behaviors identified in the previous study could successfully differentiate the two groups.

Two behaviors — leaving an object on the platform and pacing back and forth from the edge of the platform — identified 24 percent of the suicide attempters, with no false positives. Other behaviors were present in both groups; however, observing more than one of these behaviors at a time was associated with a suicide attempt.

The researchers concluded that automated computer monitoring of these indicated behaviors signals video surveillance may "identify attempters with few false positives, and potentially save lives" with appropriate intervention techniques.

More articles on health IT:
Study: 29% of STI apps distribute harmful information
IBM Security: 70% of businesses victimized by ransomware agree to pay
American Well, Concentra partner for virtual occupational medicine services

© Copyright ASC COMMUNICATIONS 2020. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.