Deploy automated CCTV/AI behaviour-detection to alert staff
#00158
Use CCTV analytics/AI to detect at-risk behaviour (loitering at platform ends, letting trains pass, trackside intrusion) and alert staff in real time — technically promising but without empirical evidence of reducing suicides in practice.
Parent issue
#00148 People in acute crisis reach platforms and trackside undetected, so few attempts are interrupted
Sustainable Development Goals
Location
Description
The proposal
Layer computer-vision analytics onto existing station/trackside CCTV to flag at-risk behaviour — prolonged loitering near the platform edge, letting multiple trains pass, trackside intrusion — and push real-time alerts to control-room and platform staff for intervention.
Why it might work
At-risk people currently pass unnoticed. Automated detection could extend limited human attention across many cameras and buy the seconds needed for a staff response.
Evidence
Detection algorithms report promising technical metrics, but there is no empirical evidence that CCTV/AI detection reduces suicides in practice. Passive surveillance presence alone has not lowered deaths.
Implementation path
Pilot on high-incidence stations with clear alert-to-response protocols; measure interventions and outcomes, not just detections; integrate with staff training — the human response is what acts.
Trade-offs and limitations
Unproven effectiveness; false positives and alert fatigue; privacy and surveillance-ethics concerns; only as good as the human response it triggers. Should be trialled and evaluated, not rolled out as proven.
Sub-issues
0Case studies
0