arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.07978cs.CL

跨语言人格驱动的数据合成用于鲁棒多模态认知衰退检测

Cross-Linguistic Persona-Driven Data Synthesis for Robust Multimodal Cognitive Decline Detection

Engineering Research Center of Intelligent Theranostics Technology and Instruments, Ministry of Education, School of Biomedical Engineering and Informatics, Nanjing Medical University（智能诊疗技术与仪器工程研究中心、教育部、生物医学工程与信息学院、南京医科大学）；Institute of Biomedical Engineering, University of Oxford（生物医学工程研究所、牛津大学）；Institute of High Performance Computing, Agency for Science, Technology and Research (A*STAR)（高性能计算研究所、科技研究局（A*STAR））；Department of Neurology, National Neuroscience Institute, Singapore, 308433, Singapore（神经病学部、新加坡国家神经科学研究所）；Duke-NUS Medical School, Singapore, 169857, Singapore（新加坡国立大学医学院）

Rui Feng, Zhiyao Luo, Liuyu Wu, Wei Wang, Yuting Song, Yong Liu, Kok Pin Ng, Jianqing Li, Xingyao Wang

备注：
18 pages, 7 figures, 6 tables
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

SynCog通过跨语言人格驱动数据合成与推理链微调，提升多模态模型在认知衰退检测中的诊断性能和跨语言泛化能力。

摘要：

基于语音的数字生物标记物代表了大规模、非侵入性地早期识别轻度认知障碍（MCI）的前沿。然而，开发稳健的诊断模型仍然受到急性临床数据稀缺和缺乏可解释推理的阻碍。当前的解决方案经常在跨语言泛化方面遇到困难，并且无法提供临床信任所需的透明理由。为了解决这些障碍，我们引入了SynCog，一种新的框架，整合了可控的零样本多模态数据合成与推理链（CoT）推理微调。具体来说，SynCog模拟了具有不同认知特征的多样化虚拟主体，以有效缓解临床数据稀缺。这种生成范式使临床语料库能够在不同语言中快速、零样本扩展，有效绕过低资源设置中的数据瓶颈，并增强多模态大语言模型（MLLMs）的诊断性能。利用这个合成数据集，我们使用CoT推理策略微调了一个基础多模态骨干网络，使模型能够明确表达诊断推理过程，而不是依赖黑箱预测。在ADReSS和ADReSSo基准上的大量实验表明，通过合成表型补充有限的临床数据可获得具有竞争力的诊断性能，分别达到宏F1分数80.67%和78.46%，优于当前基线模型。此外，对独立真实世界汉语队列（CIR-E）的评估显示了稳健的跨语言泛化能力，达到了宏F1分数48.71%。这些发现是朝着提供临床可信且语言包容的认知评估工具迈出的关键一步。

英文摘要：

Speech-based digital biomarkers represent a scalable, non-invasive frontier for the early identification of Mild Cognitive Impairment (MCI). However, the development of robust diagnostic models remains impeded by acute clinical data scarcity and a lack of interpretable reasoning. Current solutions frequently struggle with cross-lingual generalization and fail to provide the transparent rationales essential for clinical trust. To address these barriers, we introduce SynCog, a novel framework integrating controllable zero-shot multimodal data synthesis with Chain-of-Thought (CoT) deduction fine-tuning. Specifically, SynCog simulates diverse virtual subjects with varying cognitive profiles to effectively alleviate clinical data scarcity. This generative paradigm enables the rapid, zero-shot expansion of clinical corpora across diverse languages, effectively bypassing data bottlenecks in low-resource settings and bolstering the diagnostic performance of Multimodal Large Language Models (MLLMs). Leveraging this synthesized dataset, we fine-tune a foundational multimodal backbone using a CoT deduction strategy, empowering the model to explicitly articulate diagnostic thought processes rather than relying on black-box predictions. Extensive experiments on the ADReSS and ADReSSo benchmarks demonstrate that augmenting limited clinical data with synthetic phenotypes yields competitive diagnostic performance, achieving Macro-F1 scores of 80.67% and 78.46%, respectively, outperforming current baseline models. Furthermore, evaluation on an independent real-world Mandarin cohort (CIR-E) demonstrates robust cross-linguistic generalization, attaining a Macro-F1 of 48.71%. These findings constitute a critical step toward providing clinically trustworthy and linguistically inclusive cognitive assessment tools for global healthcare.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

↑