Profiling Patient Transcript Using Large Language Model Reasoning Augmentation for Alzheimer's Disease Detection

TL;DR

The paper addresses early detection of Alzheimer's disease from spontaneous speech by moving from utterance-level features to patient-level transcript profiling. It introduces an LLM reasoning augmentation framework that prompts an LLM to extract a linguistic deficit profile and summarizes it into embeddings that augment an ALBERT-based AD detector. The method achieves substantial accuracy and F1 improvements on the ADReSS Challenge dataset and offers enhanced interpretability through explicit linguistic attributes. The work highlights the potential of combining global transcript perspectives with reasoning-enabled representations to improve detection and clinical explainability, with implications for deployment and future extensions to other conditions.

Abstract

Alzheimer's disease (AD) stands as the predominant cause of dementia, characterized by a gradual decline in speech and language capabilities. Recent deep-learning advancements have facilitated automated AD detection through spontaneous speech. However, common transcript-based detection methods directly model text patterns in each utterance without a global view of the patient's linguistic characteristics, resulting in limited discriminability and interpretability. Despite the enhanced reasoning abilities of large language models (LLMs), there remains a gap in fully harnessing the reasoning ability to facilitate AD detection and model interpretation. Therefore, we propose a patient-level transcript profiling framework leveraging LLM-based reasoning augmentation to systematically elicit linguistic deficit attributes. The summarized embeddings of the attributes are integrated into an Albert model for AD detection. The framework achieves 8.51\% ACC and 8.34\% F1 improvements on the ADReSS dataset compared to the baseline without reasoning augmentation. Our further analysis shows the effectiveness of our identified linguistic deficit attributes and the potential to use LLM for AD detection interpretation.

Figures (2)

• Figure 1: Overview of our proposed LLM reasoning augmented AD detection framework. The transcripts from each participant and a designed prompt are used for linguistic attribute detection and summary using an LLM. The attributes and summary are transformed as embeddings to concatenate with the latent layer of the BERT backbone model for AD and HC sentence classification. Finally, for each participant's session, we conducted a majority vote on the prediction results of each sentence from the dialogue and form the final prediction $\hat{y}$.
• Figure 2: Prompt designed to instruct LLM for personal linguistic profiling.