LLMs show promise in diagnosing dementia and depression from interviews

Researchers have explored the use of open-weights Large Language Models (LLMs) to assess dementia and depression severity from clinical interview transcripts. The study compared three LLMs—Mistral 3.1, DeepHermes, and Qwen3—using both zero-shot prediction and feature extraction for Support Vector Regression. Results indicated that LLMs are effective for predicting depression severity in a zero-shot setting, achieving a mean absolute error of 0.60. Dementia assessment, however, saw significant improvement with structured feature extraction, reducing errors by up to 35% compared to zero-shot methods. AI

IMPACT LLMs demonstrate potential for automated screening of neuropsychiatric disorders, aiding in differential diagnosis and reducing diagnostic errors.

RANK_REASON The cluster contains an academic paper detailing research findings on LLM application for medical assessment. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• DeepHermes
• Global Depression Scale (GDS-D)
• Global Deterioration Scale (GDS)
• Hugging Face
• Mistral 3.1
• Qwen3
• Support Vector Regression

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →