Alzheimer's disease detection based on large language model prompt engineering

TL;DR

This paper addresses non-invasive, speech-based screening for Alzheimer's disease by leveraging large language models with prompt engineering. It systematically compares BERT-based prompt learning to LLAMA2-7B approaches employing Prompt Learning with LoRA, Prompt Tuning, and Conditional Learning on the ADReSS2020 dataset. The LLAMA2-7B Prompt Learning method achieves the best cross-validation accuracy of 0.8131, outperforming the BERT baseline of 0.7685, while other LLAMA2 strategies show mixed results due to data and optimization constraints. The work demonstrates a scalable, resource-efficient screening pipeline with practical clinical implications and outlines concrete directions for improving prompt-based strategies and expanding to additional languages.

Abstract

In light of the growing proportion of older individuals in our society, the timely diagnosis of Alzheimer's disease has become a crucial aspect of healthcare. In this paper, we propose a non-invasive and cost-effective detection method based on speech technology. The method employs a pre-trained language model in conjunction with techniques such as prompt fine-tuning and conditional learning, thereby enhancing the accuracy and efficiency of the detection process. To address the issue of limited computational resources, this study employs the efficient LORA fine-tuning method to construct the classification model. Following multiple rounds of training and rigorous 10-fold cross-validation, the prompt fine-tuning strategy based on the LLAMA2 model demonstrated an accuracy of 81.31\%, representing a 4.46\% improvement over the control group employing the BERT model. This study offers a novel technical approach for the early diagnosis of Alzheimer's disease and provides valuable insights into model optimization and resource utilization under similar conditions. It is anticipated that this method will prove beneficial in clinical practice and applied research, facilitating more accurate and efficient screening and diagnosis of Alzheimer's disease.

Figures (2)

• Figure 1: Schematic diagram of the fine-tuning method based on the LLAMA2 model using prompt in combination with LORA.
• Figure 2: 10-fold cross-validation accuracy results.