Every 40 seconds, someone in the world commits suicide. In Israel alone, 500 people take their own lives in an average year. In 2003, the International Association for Suicide Prevention, in conjunction with the World Health Organization (WHO), chose September 10 to be World Suicide Prevention Day, with the aim of focusing attention on the tragic issue, reducing stigma, and raising awareness.
But the death toll continues. What can be done that hasn’t been tried already, to find these people, and offer them help in time?
In retrospect, those close to a suicide victim often say there were signs. But then and there, they couldn’t recognize those signs”
Dr. Shiri Daniels, ERAN professional director
“We have the tools to try and help,” said Dr. Shiri Daniels, the professional director at ERAN, Israel’s Emotional First Aid service. “But even professional psychologists struggle to recognize suicidal ideation ahead of time. In retrospect, those close to a suicide victim often say there were signs. But then and there, they couldn’t recognize those signs.”
Can AI tools help put together the signs that slip under the radar? “People present more of themselves on social media than they realize,” said Prof. Avigdor Gal from the Faculty of Data and Decision Sciences at the Technion-Institute of Technology in Haifa. “In a three-day hackathon, 70 of its students tackled the challenge, looking for ways to recognize suicidal ideation and offer assistance. The hackathon is always dedicated to helping with a specific social issue. We cannot be blind to the society around us, what is done with the tools we develop, and how the data we collect is used. Our values must be part of the work we do.”
Hackathons are an opportunity for students to gain hands-on experience tackling the kind of challenges they will face in their future careers. In that vein, Bachelor’s students from all study tracks (industrial engineering and management; information system engineering; data science and engineering) at the faculty take part in the department’s annual hackathon.
This year, the students split into 16 teams, each comprising students from different years and tracks. Each team was given $50 to use on GPT API, offering students an opportunity to get familiar with the tool and to learn budget management.
The data used in the hackathon were part of the research of doctoral student Shir Lisak and master of science student Ilanit Sobol, who both work on recognizing suicidal behaviors online under the supervision of Prof. Roi Reichart in the Natural Language Processing (NLP) lab at the faculty.
Results from the competition
First place went to Ziv Barzilay, Liad Domb, Omri Lazover, and Jonathan Wolloch who proposed a system that would use information gathered from social media to target the inner circle of individuals displaying suicidal ideation with ads and banners related to the subject. The team members hope that those closest to those at risk would be more aware of and sensitive to the signs that precede acts of self-harm and would be best positioned to help.
Two teams tied for second place – Idan Horowitz, Lian Fichman, Shir Geisler, and Ariel Cohen used facial expression identification and NLP tools to analyze the video and transcription of posted videos, providing real-time insights to mental health professionals about the emotional well-being and suicide risk of their patients, aiding their diagnosis.
Ariel Novominsky, Alexander Freydin and Vladislav Comantany tried to characterize the emotional process a person undergoes before deciding to commit suicide and proposed a quantitative measurement method for the abstract concept, an emotional process, in order to allows for mathematical analysis.
The Data for Good Israel, ERAN, Gila’s Way and Bishvil Ha’hayim (Path to Life) associations enthusiastically joined the event and gave the students the information and motivation to develop solutions that may save lives in the future. With the help of these organizations, the students were able to approach the painful subject of suicide with as much information as possible.
They could also begin to discuss this difficult topic in an open and sensitive manner. Being alert to the signs of people suffering from suicidal ideation and signs of deterioration before taking action could, the organizations hoped, be useful beyond the narrow scope of the Hackathon.