ABSTRACT

For this observational cross-sectional study, different modalities of verbal fluency tasks (VFTs) were compared between 143 participants: 35 cognitively healthy controls (CHCs), 71 mild cognitive impairment (MCI) and 37 mild Alzheimer's disease (AD) patients. Binomial logistic regression models were defined to identify VFT variables associated with MCI and AD, with respect to CHC. The results showed that the best errors/repetitions variable associated with MCI and AD was the phonemic task, and with every error the odds of being in the MCI group increased 9.9 times and 12.2 times in AD group, accompanied by high accuracy values (MCI: AUC = 0.824, sensitivity = 0.676, specificity = 0.943; AD: AUC = 0.883, sensitivity = 0.784, specificity = 0.943). The results suggest that, in addition to solely register raw scores, a simple counting of errors and repetitions during VFT can offer valuable clues in detecting MCI and AD, especially in the phonemic task.

摘要

对于这项观察性横断面研究，比较了 143 名参与者的不同形式的语言流畅性任务 (VFT)：35 名认知健康对照组 (CHC)、71 名轻度认知障碍 (MCI) 和 37 名轻度阿尔茨海默病 (AD) 患者。二项逻辑回归模型被定义为识别与 MCI 和 AD 相关的 VFT 变量，关于 CHC。结果表明，与 MCI 和 AD 相关的最佳错误/重复变量是音素任务，每出现一个错误，在 MCI 组中的几率增加 9.9 倍，在 AD 组中增加 12.2 倍，并伴随着高精度值（MCI： AUC = 0.824，灵敏度 = 0.676，特异性 = 0.943；AD：AUC = 0.883，灵敏度 = 0.784，特异性 = 0.943）。结果表明，除了只记录原始分数外，