Seoul City uses AI tech to prevent suicide attempts on Han River bridges

The Korea Times

The Korea Times

National

Sun, September 24, 2023 | 13:42

Seoul City uses AI tech to prevent suicide attempts on Han River bridges

Posted : 2021-12-16 13:43

Updated : 2021-12-16 18:14

Monitors at the integrated surveillance camera control center of the Ttukseom Water Rescue Brigade display footage of bridges across the Han River in Seoul, in this provided photo. Courtesy of Seoul Metropolitan Government

By Bahk Eun-ji

Korea bears the disgrace of having one of the world's highest suicide rates, topping the ranking among OECD member nations with 23.5 out of 100,000 people killing themselves in 2020.

The central and local governments have taken various suicide prevention measures, and the Seoul Metropolitan Fire and Disaster Headquarters has recently come up with a system using artificial intelligence (AI) technology along Han River bridges, where about 500 suicide attempts are made every year.

The AI-based technology, developed in cooperation with the Seoul Institute of Technology (SIT) since April 2020, has undergone deep learning of patterns of behavior of people attempting suicide by jumping off the bridges, such as wandering on an area of a bridge for a period of minutes, as well as other data, including information on rescue teams' dispatch history, surveillance camera recordings and conversations through suicide hotlines.

When the cameras detect a person showing such behaviors, the system immediately sends a signal to rescue teams near the bridge so they can arrive at the scene faster to intervene or carry out rescue operations.

"The most important thing in rescuing people who are trying to commit suicide is that they must be stopped before they're in the water. Once that happens, the survival rate drops below 50 percent," an official of the disaster headquarters said.

The official said the system helps rescue team members detect dangerous situations among hundreds of surveillance feeds.

"We monitor 572 CCTV cameras in the control center, so it is not easy for a handful of workers to catch everything. But now the AI system selects footage of a person showing suspicious behavior and sounds an alarm, so that rescue teams can respond much faster," he said.

Ttukseon Water Rescue Brigade / Courtesy of Seoul Metropolitan Government

There are 28 bridges crossing the Han River. Before the AI system, there have been other efforts to prevent suicides from the bridges, such as safety fences equipped with laser detection sensors.

In particular, the safety fence installed on Mapo Bridge senses if a person is gripping the guardrail and sends a signal to rescue teams. Cameras installed on Seogang Bridge send alarms if they detect a human body under the bridge.

With these various technologies combined, 96.63 percent of people attempting suicide from the bridges have been stopped or rescued over the past five years.

In addition to those existing prevention systems, if the AI detection system is put in use, the rescue rate is expected to be close to 100 percent, the official said.

With the new system, the Ttukseom Water Rescue Brigade will be the integrated control center for Han River bridges for more efficient control. Previously four brigades ― Yeouido, Banpo, Gwangnaru and Ttukseom ― had their own respective monitoring and dispatch systems.

The AI system is undergoing a trial period until the end of this month. "When a little more data is accumulated and this system is well established, it is expected to be significantly helpful in preventing suicide," a city government official said.

"The city government has been looking for a solution as the number of people who commit suicide in the Han River is rising due to the prolonged COVID-19 pandemic and increasing financial difficulties," he said.

Ko In-seok, head of the SIT, said, "The system is an example of cooperation between the city government and the research institute to minimize human casualties caused by suicide attempts at the bridges of the Han River. We will continue to conduct practical research for the safety of citizens by using data science analysis technology."

ejb@koreatimes.co.kr