Can CCTV identify people in public transit stations who are at risk of attempting suicide? An analysis of CCTV video recordings of attempters and a comparative investigation

Abstract

BackgroundSuicides incur in all public transit systems which do not completely impede access to tracks. We conducted two studies to determine if we can reliably identify in stations people at risk of suicide in order to intervene in a timely manner. The first study analysed all CCTV recordings of suicide attempters in Montreal underground stations over 2 years to identify behaviours indicating suicide risk. The second study verified the potential of using those behaviours to discriminate attempters from other passengers in real time.Methods First study: Trained observers watched CCTV video recordings of 60 attempters, with 2–3 independent observers coding seven easily observable behaviours and five behaviours requiring interpretation (e.g. “strange behaviours,” “anxious behaviour”). Second study: We randomly mixed 63 five-minute CCTV recordings before an attempt with 56 recordings from the same cameras at the same time of day, and day of week, but when no suicide attempt was to occur. Thirty-three undergraduate students after only 10 min of instructions watched the recordings and indicated if they observed each of 13 behaviours identified in the First Study.Results First study: Fifty (83%) of attempters had easily observable behaviours potentially indicative of an impending attempt, and 37 (61%) had two or more of these behaviours. Forty-five (75%) had at least one behaviours requiring interpretation. Twenty-two witnesses attempted to intervene to stop the attempt, and 75% of attempters had behaviours indicating possible ambivalence (e.g. waiting for several trains to pass; trying to get out of the path of the train). Second study: Two behaviours, leaving an object on the platform and pacing back and forth from the yellow line (just before the edge of the platform), could identify 24% of attempters with no false positives. The other target behaviours were also present in non-attempters. However, having two or more of these behaviours indicated a likelihood of being at risk of attempting suicide.ConclusionsWe conclude that real time observations of CCTV monitors, automated computer monitoring of CCTV signals, and/or training of drivers and transit personnel on behavioural indications of suicide risk, may identify attempters with few false positives, and potentially save lives.

Behaviours preceding suicides at railway and underground locations: a multimethodological qualitative approach

“… 9 10 Only one publication that focused on Canadian railway locations has used CCTV data to establish possible identifiable behaviour in the moments preceding a suicide. 8 There are no publications which incorporate an analysis of CCTV data of people who have died by rail suicide with first-hand accounts of suicidal behaviour at railway locations, from the perspectives of those who have survived and/or have witnessed such behaviour at railway locations.…”

The Use of Closed-Circuit Television and Video in Suicide Prevention: Narrative Review and Future Directions

“…In one study, over the course of 2 experiments, the authors attempted to identify behaviors indicative of a subsequent suicide attempt at metro stations, and tested how well these behaviors predict an attempt [ 8 ]. In the first experiment, the authors identified key behaviors by having 2-3 observers code both easily observable behaviors (eg, standing near the edge of the platform) and those behaviors requiring interpretation (eg, looking depressed).…”

“…Another study [ 9 ] expanded on the findings from the aforementioned study [ 8 ] and used semistructured interviews and questionnaires administered to railway staff in addition to CCTV footage to identify these potential indicator behaviors. The process identified 5 main behaviors: station hopping and platform switching, limited contact with people, allowing trains to pass by, where they stood on the platform, and repetitive behaviors such as walking up and down the length of the platform.…”

“…Newer approaches utilize automated systems, such as using thermal motion sensors or defining a region of interest in CCTV scenes, while others utilize computer vision to detect behaviors indicative of a suicide attempt by hanging [ 20 , 33 , 36 ]. Finally, others have examined crisis behaviors as a starting point to develop interventions that are able to reach out even earlier [ 8 , 9 ].…”

“…A range of suicide prevention initiatives in public places have been evaluated, with the use of CCTV cameras being proposed as a means to increase the likelihood of a third-party intervention [ 6 ], for example, by initiating a police call-out after climbing a safety fence [ 7 ]. Studies from the metro/underground railway settings have retrospectively analyzed behaviors which precede a suicide attempt, with observable behaviors being identified [ 8 , 9 ]. These studies have further suggested that it may be possible to automatically detect behaviors prior to a suicide attempt, potentially allowing for earlier intervention and the interruption of a suicide attempt.…”