Software for improving suicide prevention
If you work for a suicide prevention chatline, a mental block can mean the difference between life and death, says Salim Salmi, who developed software to assist staff dealing with callers.
For his graduation project in the Faculty of Electrical Engineering, Mathematics and Computer Science, Salim Salmi developed a support system for chatline staff. This computer program analyses thousands of previous conversations and uses natural language processing techniques (a form of AI) to identify similar scenarios that staff members can use to get ideas to keep difficult conversations going.
Salmi used the database of the 113 suicide prevention helpline. With his colleague, Saskia Mérelle, from 113, he sifted through seven months of chats (from March to September 2018). “We started by filtering the chat data, removing anything with fewer than 20 interactions. We used 17,773 conversations in total.” The study results were recently published in the Journal of Medical Internet Research.
The program still requires further improvement. Chatline staff said that they preferred to enlist the support of more experienced colleagues in complex situations. However, this is not always possible in practice. Salmi is now working as a doctoral candidate at the Amsterdam institute for mathematics and computer Science CWI and is improving the software.