Towards Within-Class Variation in Alzheimer's Disease Detection from Spontaneous Speech

2024-09-22

arXiv.org

AI Summary

Alzheimer’s disease (AD) speech detection faces three key challenges: substantial intra-class variability in cognitive impairment severity, absence of fine-grained severity labels, and instance-level class imbalance—leading to poor generalization of conventional binary classifiers. To address these, we depart from the binary classification paradigm and propose Soft Target Distillation (SoTD) to model AD severity as a continuous spectrum, coupled with Instance-level Rebalancing (InRe) to mitigate severity distribution shift. Our method integrates multi-model knowledge distillation, severity-aware dynamic weighted sampling, and end-to-end speech representation learning. Evaluated on the ADReSS and ADReSSo benchmarks, our approach achieves significant accuracy improvements over baselines. SoTD enhances discriminative consistency across severity levels, while InRe effectively suppresses overfitting. Experimental results demonstrate that our framework robustly models real-world clinical heterogeneity in AD progression.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Natural Language Processing: Speech
• Search and Optimization: Adversarial Search

Application Category

• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Alzheimer's Disease (AD) detection has emerged as a promising research area that employs machine learning classification models to distinguish between individuals with AD and those without. Unlike conventional classification tasks, we identify within-class variation as a critical challenge in AD detection: individuals with AD exhibit a spectrum of cognitive impairments. Given that many AD detection tasks lack fine-grained labels, simplistic binary classification may overlook two crucial aspects: within-class differences and instance-level imbalance. The former compels the model to map AD samples with varying degrees of impairment to a single diagnostic label, disregarding certain changes in cognitive function. While the latter biases the model towards overrepresented severity levels. This work presents early efforts to address these challenges. We propose two novel methods: Soft Target Distillation (SoTD) and Instance-level Re-balancing (InRe), targeting two problems respectively. Experiments on the ADReSS and ADReSSo datasets demonstrate that the proposed methods significantly improve detection accuracy. Further analysis reveals that SoTD effectively harnesses the strengths of multiple component models, while InRe substantially alleviates model over-fitting. These findings provide insights for developing more robust and reliable AD detection models.

Problem

• Addresses within-class variation in Alzheimer's Disease detection
• Mitigates within-class heterogeneity and instance-level imbalance issues
• Proposes methods to improve detection performance using cognitive scores

Innovation

• Sample score estimator for soft scores
• Soft Target Distillation (SoTD) method
• Instance-level Re-balancing (InRe) technique