目的：探究空间记忆检测（spatial memory test，SMT）在中国老年人群中识别轻度认知障碍（mild cognitive impairment，MCI）的效能，并与蒙特利尔认知评估（Montreal cognitive assessment，MoCA）进行比较，为中国个体化筛查MCI提供依据。方法：本研究数据来自上海市老年人脑健康队列研究，选取自2023年1月至2024年1月期间连续入组的415名社区老年人，包括认知正常者（cognitively normal，NC）88例、主观认知下降（subjective cognitive decline， SCD）者188例和MCI患者139例。所有受试者均完成SMT与MoCA评估。以SCD、NC为对照组，以MCI为病例组，采用受试者操作特征曲线（receiver operating characteristic， ROC）曲线分析比较2种工具的筛查效能，进一步探讨SMT评分与不同神经心理学检验间的相关性，通过分层及交叉分析探讨年龄和教育水平对其筛查效能的影响。结果：SMT与MoCA均能较好地识别MCI，且SMT展现出更优的整体筛查效能。在区分SCD与MCI时，MoCA的ROC曲线的曲线下面积（area under the curve， AUC）略高于SMT（0.925比0.885），但SMT在国际常用截断值下（≤8分）具备更高的筛查特异度（91.06%）和阳性预测值（84.89%），整体准确率达（80.0%），优于不同截断值下MoCA的综合表现。相关性分析发现，SMT不仅与MoCA总分呈正相关（r=0.612，P<0.001），且与延迟记忆、视空间推理这些AD早期敏感认知领域密切相关（P均<0.01）。分层分析显示，在年龄≥75岁人群中SMT、MoCA区分SCD与MCI的AUC无差异（0.917比0.910，P=0.802）；在教育程度>9年的人群中两者区分SCD与MCI效能相当（AUC：0.924比0.904，P=0.358）；交叉分析进一步表明，仅在教育程度>9年且年龄≥75岁的亚组中，MoCA区分SCD与MCI（AUC：0.942比0.864， P<0.05）及NC与MCI（AUC：0.99比0.90， P<0.05）的效能略优于SMT。结论：SMT作为一种数字化早期认知障碍的筛查工具，其筛查MCI的整体效能优于MoCA，且具有高特异度、操作便捷等优势，更适用于社区大规模筛查及高龄人群快速初筛。

Objective To evaluate the screening performance of the Spatial Memory Test (SMT) in identifying mild cognitive impairment (MCI) among the elderly population in China, and to compare it with the Montreal Cognitive Assessment (MoCA), in order to provide a foundation for individualized MCI screening in China. Methods The data for this study were derived from the Shanghai Elderly Brain Health Cohort Study, including 415 community-dwelling older adults consecutively enrolled from January 2023 to January 2024, comprising 88 cognitively normal (NC) individuals, 188 with subjective cognitive decline (SCD), and 139 individuals with mild cognitive impairment (MCI). All participants completed SMT and MoCA assessments. SCD and NC were served as the control groups and MCI as the case group, and receiver operating characteristic (ROC) curve analysis was employed to compare the screening performance of the two tools. Furthermore, the correlation between SMT scores and various neuropsychological tests was explored, and stratified and interaction analyses were conducted to examine the effects of age and education level on the screening performance of SMT. Results Both SMT and MoCA effectively identified MCI, with SMT demonstrating superior overall screening performance. When distingui-shing SCD from MCI, the area under the receiver operating characteristic curve (AUC) of MoCA was slightly higher than that of SMT (0.925 vs. 0.885). However, SMT showed higher specificity (91.06%) and positive predictive value (84.89%) at the commonly used international cutoff point (≤8), with an overall accuracy of 80.0%, outperforming MoCA across different cutoff values. Correlation analysis indicated that SMT was not only positively correlated with the total MoCA score (r=0.612, P<0.001), but also closely associated with cognitive domains sensitive to early AD, such as delayed recall and visuospatial reasoning (all P<0.01). Stratified analysis revealed that among individuals aged ≥75 years, the AUC for distinguishing SCD from MCI was comparable between SMT and MoCA (0.917 vs. 0.910, P=0.802); among those with education >9 years, the performance of the two tests in distinguishing SCD from MCI was similar (AUC: 0.924 vs. 0.904, P=0.358). Interaction analysis further indicated that only in the subgroup with education >9 years and age ≥75 years did MoCA slightly outperform SMT in distinguishing SCD from MCI (AUC: 0.942 vs. 0.864, P<0.05) and NC from MCI (AUC: 0.99 vs. 0.90, P<0.05). Conclusion As a digital screening tool for early cognitive impairment, SMT demonstrates superior overall performance compared to MoCA in screening for MCI, with advantages of high specificity and ease of use, and is more suitable for large-scale community screening and rapid preliminary assessment in the elderly population.