Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Integrating Pause Information with Word Embeddings in Language Models for Alzheimer's Disease Detection from Spontaneous Speech

Yu Pu, Wei-Qiang Zhang

TL;DR

This work tackles early Alzheimer's disease detection from spontaneous speech by leveraging pauses as temporal cues. It introduces a pause-embedding mechanism that encodes word durations and inter-word pauses, and integrates these embeddings into a BERT-based language model. A two-stage training regime uses GigaSpeech for pretraining the pause representations and ADReSSo for task-specific fine-tuning, achieving a top-1 accuracy of 83.1% on ADReSSo. The results demonstrate that pause information is a valuable non-invasive biomarker for AD and suggests broader applicability to neurodegenerative disease detection.

Abstract

Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline and memory loss. Early detection of AD is crucial for effective intervention and treatment. In this paper, we propose a novel approach to AD detection from spontaneous speech, which incorporates pause information into language models. Our method involves encoding pause information into embeddings and integrating them into the typical transformer-based language model, enabling it to capture both semantic and temporal features of speech data. We conduct experiments on the Alzheimer's Dementia Recognition through Spontaneous Speech (ADReSS) dataset and its extension, the ADReSSo dataset, comparing our method with existing approaches. Our method achieves an accuracy of 83.1% in the ADReSSo test set. The results demonstrate the effectiveness of our approach in discriminating between AD patients and healthy individuals, highlighting the potential of pauses as a valuable indicator for AD detection. By leveraging speech analysis as a non-invasive and cost-effective tool for AD detection, our research contributes to early diagnosis and improved management of this debilitating disease.

Integrating Pause Information with Word Embeddings in Language Models for Alzheimer's Disease Detection from Spontaneous Speech

TL;DR

This work tackles early Alzheimer's disease detection from spontaneous speech by leveraging pauses as temporal cues. It introduces a pause-embedding mechanism that encodes word durations and inter-word pauses, and integrates these embeddings into a BERT-based language model. A two-stage training regime uses GigaSpeech for pretraining the pause representations and ADReSSo for task-specific fine-tuning, achieving a top-1 accuracy of 83.1% on ADReSSo. The results demonstrate that pause information is a valuable non-invasive biomarker for AD and suggests broader applicability to neurodegenerative disease detection.

Abstract

Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline and memory loss. Early detection of AD is crucial for effective intervention and treatment. In this paper, we propose a novel approach to AD detection from spontaneous speech, which incorporates pause information into language models. Our method involves encoding pause information into embeddings and integrating them into the typical transformer-based language model, enabling it to capture both semantic and temporal features of speech data. We conduct experiments on the Alzheimer's Dementia Recognition through Spontaneous Speech (ADReSS) dataset and its extension, the ADReSSo dataset, comparing our method with existing approaches. Our method achieves an accuracy of 83.1% in the ADReSSo test set. The results demonstrate the effectiveness of our approach in discriminating between AD patients and healthy individuals, highlighting the potential of pauses as a valuable indicator for AD detection. By leveraging speech analysis as a non-invasive and cost-effective tool for AD detection, our research contributes to early diagnosis and improved management of this debilitating disease.
Paper Structure (17 sections, 4 figures, 3 tables)

This paper contains 17 sections, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Speech recognition with timestamps
  4. Pauses and word durations extraction
  5. Integration of pause information into language models
  6. Model training and fine-tuning
  7. Experiment
  8. Dataset
  9. GigaSpeech
  10. ADReSS
  11. ADReSSo
  12. Experimental Setting
  13. Results
  14. Pause Prediction Task
  15. Alzheimer's Detection Task
  16. ...and 2 more sections

Figures (4)

  • Figure 1: General flowgram of our AD detection system. Our main innovation lies in the part marked with a pentagram, which integrates pause information with the language model.
  • Figure 2: The detailed process of embedding calculation.
  • Figure 3: The distribution of pause durations in the ADReSSo dataset, with pause duration between 0.8s and 3s amplified.
  • Figure 4: Results of the pause prediction task.