An unsupervised adaptive-RAG pipeline that retrieves a user's Reddit posts for each BDI-II item and has an LLM score the item matches or outperforms supervised benchmarks on eRisk depression and other mental-health screening tasks.
Francesco Denti, Diego Doimo, Alessandro Laio, and Antonietta Mira
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Are LLMs effective psychological assessors? Leveraging adaptive RAG for interpretable mental health screening through psychometric practice
An unsupervised adaptive-RAG pipeline that retrieves a user's Reddit posts for each BDI-II item and has an LLM score the item matches or outperforms supervised benchmarks on eRisk depression and other mental-health screening tasks.