Richard Amélie B, Lelandais Manon, Reilly Karen T, Jacquin-Courtois Sophie
Université Montpellier Paul-Valéry, Praxiling UMR 5267, France.
Université de Lyon, Lyon Neuroscience Research Centre, France.
J Speech Lang Hear Res. 2024 Dec 9;67(12):4714-4733. doi: 10.1044/2024_JSLHR-24-00274. Epub 2024 Nov 15.
This systematic review covers the current stage of research on subtle cognitive impairment with connected speech. It aims at surveying the linguistic features in use to single out those that can best identify patients with mild neurocognitive disorders (mNCDs), whose cognitive changes remain underdiagnosed.
We followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines and proposed a full definition of features for the analysis of speech features. Fifty-one studies met the inclusion criteria. Most of them focused on age-related progressive diseases and included fewer than 30 subjects.
A total of 384 features labeled with 335 different names was retrieved, yielding various results in discriminating individuals with mNCDs from controls.
This finding highlights the need for harmonized labels to further investigate mNCDs with linguistic markers. We suggest two different ways of assessing a feature's reliability. We also point out potential methodological issues that remain to be resolved, along with recommendations for reproducible research in the field.
本系统评价涵盖了与连贯言语相关的轻度认知障碍的当前研究阶段。其目的是调查所使用的语言特征，以找出那些最能识别轻度神经认知障碍（mNCD）患者的特征，这些患者的认知变化仍未得到充分诊断。
我们遵循系统评价和荟萃分析的首选报告项目指南，并提出了用于分析言语特征的特征的完整定义。51项研究符合纳入标准。其中大多数研究集中在与年龄相关的进行性疾病上，且受试者少于30名。
共检索到384个特征，有335个不同的名称，在区分mNCD患者与对照组方面产生了各种结果。
这一发现凸显了使用统一标签来进一步研究具有语言标记的mNCD的必要性。我们提出了两种评估特征可靠性的不同方法。我们还指出了有待解决的潜在方法学问题，以及该领域可重复研究的建议。
