MInDS: Using Large Language Models to Screen for Depression

Depression is a debilitating, yet underdiagnosed mental illness due to the subjectivity of current screening and time and resource restrictions. Large language models (LLMs) can potentially address these difficulties. Using the Extended Distress Analysis Interview Corpus dataset, containing 105 interview transcripts, we propose MInDS, an automated, modular LLM inferencing pipeline, to optimize depression screening. Our results indicate that LLMs can effectively screen for depression with a 0.8 balanced accuracy. LLM inferencing with a shortened transcript can perform similarly to inferencing with the entire transcript. Our findings may aid the future development of LLMs for depression screening.