智能化认知量表与人工评估在临床人群中评估一致性的观察  

Observation of the consistency between intellectualized and manual-based cognitive assessment tools in the outpatient clinic

作　　者：邢怡[1] 秦琪 王治斌 王丹旸 李诗怡 孙悦文 靳红梅[1] 武广山 蔡龙军 王晓怡 唐毅[1] Xing Yi;Qin Qi;Wang Zhibin;Wang Danyang;Li Shiyi;Sun Yuewen;Jin Hongmei;Wu Guangshan;Cai Longjun;Wang Xiaoyi;Tang Yi(Department of Neurology,Xuanwu Hospital,Capital Medical University,Beijing 100053,China;Beijing Wise Spirit Technology Limited,Beijing 100192,China)

机构地区：[1]首都医科大学宣武医院神经内科,北京100053 [2]北京智精灵科技有限公司,北京100192

出　　处：《中华医学杂志》2024年第8期600-607,共8页National Medical Journal of China

基　　金：国家重点研发计划(2022YFC3602600);国家自然科学基金(82220108009,81970996)

摘　　要：目的对蒙特利尔认知评估量表(MoCA)与简明智力状态检查量表(MMSE)进行智能化(即i-MoCA/i-MMSE),并在临床人群中通过比较智能化量表与人工施测量表的一致性验证其效度。方法前瞻性纳入2023年2至10月在首都医科大学宣武医院门诊就诊、主诉为认知功能下降的患者,共入组88例,年龄(66.82±11.37)岁,男30例,女58例。采用一致性研究设计,每例受试者需要分别完成人工施测的MoCA和MMSE,以及i-MoCA和i-MMSE。为控制练习效应的影响,两种测评需间隔2周完成。人工施测和智能化量表的完成顺序在受试者间进行随机。采用一致性相关系数(CCC)来计算智能化和人工施测量表的可靠性,通过协变量方差分析比较智能化和人工施测量表的差异,通过受试者工作特征曲线计算特异度与敏感度分析智能化量表的评估准确性。结果智能化量表和人工施测之间存在较高的一致性(CCC_(MoCA)=0.87,CCC_(MMSE)=0.83)。在控制了基本人口学信息后,智能化量表和人工施测的得分差异无统计学意义(P>0.05)。以人工施测作为标准,i-MoCA诊断认知障碍的准确率达94.3%(灵敏度=94.6%,特异度=78.1%),i-MMSE诊断认知障碍的准确率达94.9%(灵敏度=94.9%,特异度=77.6%)。除此以外,同样在两种量表的多项子功能之间发现了较高一致性(CCC_(MoCA)=0.32~0.78;CCC_(MMSE)=0.54~0.79)。结论智能化量表无论在整体认知功能还是子功能上都和人工施测具有较高的一致性与诊断准确性。Objective The intellectualized versions of the Montreal Cognitive Assessment Scale(MoCA)and the Mini-mental State Examination(MMSE)(i-MoCA/i-MMSE)were developed.The validity of this system was evaluated in a clinical sample through comparing with the manual-based assessments.Methods A total of 88 patients[aged(66.82±11.37)years,30 males and 58 females]were enrolled in the outpatient clinic of Xuanwu Hospital of Capital Medical University with complaints of cognitive decline,from February to October 2023.All participants completed manual-based and intellectualized assessments in a randomized order,with an interval of 2 weeks to control for the practice effect.The reliability of the intellectualized version of assessments was evaluated based on the manual-based version using the Concordance correlation coefficient(CCC).The difference between the intellectualized and the manual-based assessments was tested by the Repeated ANCOVA with demographic information controlled.The accuracy of evaluation of the i-MoCA and i-MMSE was analyzed by the Receiver Operating Characteristic(ROC)analysis.Results High concordance was observed between the intellectualized version and the manual-based assessments(CCC_(MoCA)=0.87,CCC_(MMSE)=0.83).Controlling for basic demographic information,there was no significant difference in the scores of the intellectualized version and the manual-based assessments(all P>0.05).The accuracy of i-MoCA in screening patients with cognitive impairment was 94.3%(sensitivity=94.6%,specificity=78.1%),while the accuracy of i-MMSE in screening patients with cognitive impairment was 94.9%(sensitivity=94.9%,specificity=77.6%).In addition,the majority of subdomains measured by the cognitive assessments exhibited high consistency across the intellectualized the manual-based versions(CCC_(MoCA)=0.32-0.78;CCC_(MMSE)=0.54-0.79).Conclusion Both the i-MoCA and i-MMSE showed high consistency and diagnostic accuracy with the manual-based versions in terms of overall cognitive function and subdomains.

关 键 词：认知功能异常 数字健康 蒙特利尔认知评估量表 简明智力状态检查量表 横断面研究 

分 类 号：R749.1[医药卫生—神经病学与精神病学]