MultiConAD: A Unified Multilingual Conversational Dataset for Early Alzheimer's Detection

2025-02-26

AI Summary

This study addresses two critical bottlenecks in early Alzheimer’s disease (AD) detection: (1) the frequent omission of mild cognitive impairment (MCI)—a key prodromal stage—leading to overly coarse AD/healthy binary classification; and (2) heavy reliance on monolingual (predominantly English) data, resulting in poor cross-lingual generalizability. To this end, we introduce the first unified multilingual spoken dialogue dataset for AD screening, covering English, Spanish, Chinese, and Greek, with synchronized audio and transcribed text, enabling fine-grained AD and MCI classification. Methodologically, we present the first integrated multilingual AD dialogue corpus, formalize a three-level MCI classification task, and propose a comprehensive evaluation framework incorporating sparse/dense representations, monolingual/multilingual supervised classification, and cross-lingual alignment analysis. Experiments reveal that multilingual modeling improves performance in some languages, whereas others benefit more from monolingual training—highlighting the necessity of language-specific modeling. Our work establishes a foundation for robust, generalizable cross-lingual early AD detection.

Technology Category

Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLPMachine Learning: Multimodal LearningData Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web searchWeb Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web dataEconomics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Dementia is a progressive cognitive syndrome with Alzheimer's disease (AD) as the leading cause. Conversation-based AD detection offers a cost-effective alternative to clinical methods, as language dysfunction is an early biomarker of AD. However, most prior research has framed AD detection as a binary classification problem, limiting the ability to identify Mild Cognitive Impairment (MCI)-a crucial stage for early intervention. Also, studies primarily rely on single-language datasets, mainly in English, restricting cross-language generalizability. To address this gap, we make three key contributions. First, we introduce a novel, multilingual dataset for AD detection by unifying 16 publicly available dementia-related conversational datasets. This corpus spans English, Spanish, Chinese, and Greek and incorporates both audio and text data derived from a variety of cognitive assessment tasks. Second, we perform finer-grained classification, including MCI, and evaluate various classifiers using sparse and dense text representations. Third, we conduct experiments in monolingual and multilingual settings, finding that some languages benefit from multilingual training while others perform better independently. This study highlights the challenges in multilingual AD detection and enables future research on both language-specific approaches and techniques aimed at improving model generalization and robustness.

Problem

• Multilingual Alzheimer's detection
• Early Mild Cognitive Impairment identification
• Cross-language model generalization

Innovation

• Multilingual conversational dataset unification
• Finer-grained classification including MCI
• Monolingual and multilingual training experiments