Abstract
Suicidal risk detection in adolescents is a critical challenge, yet existingmethods rely on language-specific models, limiting scalability andgeneralization. This study introduces a novel language-agnostic framework forsuicidal risk assessment with large language models (LLMs). We generate Chinesetranscripts from speech using an ASR model and then employ LLMs withprompt-based queries to extract suicidal risk-related features from thesetranscripts. The extracted features are retained in both Chinese and English toenable cross-linguistic analysis and then used to fine-tune correspondingpretrained language models independently. Experimental results show that ourmethod achieves performance comparable to direct fine-tuning with ASR resultsor to models trained solely on Chinese suicidal risk-related features,demonstrating its potential to overcome language constraints and improve therobustness of suicidal risk assessment.