MINDS CDT student among winning team of the 9th Workshop on Computational Linguistics and Clinical Psychology shared task

The cohort 4 MINDS CDT student Rudra Mutalik was part of a team that achieved an overall first place in the shared task at the 9th Workshop on Computational Linguistics and Clinical Psychology (CLPsych 2024) hosted by the prestigious 18th Conference of the European Chapter of the Association for Computational Linguistics (EACL-2024) that took place 17-22 March 2024 in Malta. The team, consisting of Dr Gyanendro Loitongbam, Junyu Mao, Rudra Mutalik & Dr Stuart Middleton, was ranked top for the evidence extraction task and second for the summarization task. This shared task involved employing large language models (LLMs) for extracting evidence and summarising findings around assessment of suicidal risk level for individuals on Reddit based on their posts.

The team's winning entry used Mixtral7bx8 and Tulu-2-DPO-70B Large Language Models. They utilised various state-of-the-art prompting techniques including in-context & chain-of-thought techniques to incite reasoning and control the response structure. Rudra and the rest of the team demonstrated the potential for LLMs in mental health practice, showing that they have the potential to synthesize diverse medical insights into meaningful, well-formed text.