Reading between the Lines: Leveraging Large Language Models for Global Dementia and Depression Assessment from Clinical Interviews

2026-06-16

AI Summary

This study addresses the diagnostic challenge posed by the symptomatic overlap between dementia and depression in older adults by proposing a parallel assessment framework based on open-source large language models (e.g., Mistral 3.1, DeepHermes, and Qwen3). Leveraging transcribed speech from standardized clinical interviews—including pause information—the work introduces a novel Global Depression Scale aligned with the Global Deterioration Scale to enable concurrent staging of affective and cognitive symptoms. Under both zero-shot inference and support vector regression–based feature extraction paradigms, the models achieve strong performance, yielding mean absolute errors of 0.60 for depression and 0.78 for dementia severity estimation; the latter represents a 35% reduction in error compared to baseline methods. Notably, model performance using automatically generated transcripts matches that of manually transcribed ones.

Abstract

Dementia and depression are the most prevalent neuropsychiatric disorders in geriatric populations, and their overlapping symptoms pose major challenges for differential diagnosis. In this study, we investigate open-weights Large Language Models (LLMs) for predicting dementia and depression severity from speech samples collected during standardized history taking interviews with 154 German-speaking subjects. We introduce an observer-based Global Depression Scale (GDS-D) aligned with the established Global Deterioration Scale (GDS), enabling parallel global staging of affective and cognitive symptoms. We compare three LLMs (Mistral 3.1, DeepHermes, Qwen3) in two settings: (1) zero-shot prediction and (2) LLM-based feature extraction for Support Vector Regression, using human and pause-enriched transcripts. Results show that LLMs effectively predict depression severity in zero-shot settings (best MAE of 0.60), while dementia assessment benefits substantially from structured feature extraction (best MAE of 0.78), reducing errors by up to 35% over zero-shot baselines. Pause-enriched transcripts achieve competitive performance with human transcriptions, demonstrating the viability of fully automatic screening pipelines for differential neuropsychiatric assessment.

Problem

• dementia
• depression
• differential diagnosis
• clinical interviews
• neuropsychiatric disorders

Innovation

• Large Language Models
• dementia assessment
• depression severity
• zero-shot prediction
• pause-enriched transcripts