University of Michigan, Ann Arbor, MI, USA.
The Michigan Alzheimer's Disease Center, Ann Arbor, MI, USA.
Assessment. 2023 Apr;30(3):847-855. doi: 10.1177/10731911211069089. Epub 2022 Jan 11.
Cognitive testing data are essential to the diagnosis of mild cognitive impairment (MCI), and computerized cognitive testing, such as the Cogstate Brief Battery, has proven helpful in efficiently identifying harbingers of dementia. This study provides a side-by-side comparison of traditional Cogstate outcomes and diffusion modeling of these outcomes in predicting MCI diagnosis. Participants included 257 older adults (160 = normal cognition; 97 = MCI). Results showed that both traditional Cogstate and diffusion modeling analyses predicted MCI diagnosis with acceptable accuracy. Cogstate measures of recognition learning and working memory accuracy and diffusion modeling variable of decision-making efficiency (drift rate) and nondecisional time were most predictive of MCI. While participants with normal cognition demonstrated a change in response caution (boundary separation) when transitioning tasks, participants with MCI did not evidence this change.
认知测试数据对于轻度认知障碍 (MCI) 的诊断至关重要，而计算机化认知测试，如 Cogstate 简短电池测试，已被证明有助于有效地识别痴呆的先兆。本研究对传统的 Cogstate 结果和这些结果的扩散建模在预测 MCI 诊断方面进行了并排比较。参与者包括 257 名老年人（160 名=正常认知；97 名=MCI）。结果表明，传统的 Cogstate 和扩散建模分析都可以以可接受的准确度预测 MCI 诊断。Cogstate 的识别学习和工作记忆准确性以及扩散建模的决策效率（漂移率）和非决策时间变量最能预测 MCI。虽然认知正常的参与者在转换任务时表现出反应谨慎（边界分离）的变化，但 MCI 参与者没有表现出这种变化。
