多状态Markov模型在阿尔茨海默病自然史中的应用  

Application of Multi-state Markov Model in the Natural History of Alzheimer's Disease

作　　者：秦瑶 韩红娟 刘龙 陈杜荣 王浩基 崔靖 白文琳 余红梅 Qin Yao;Han Hongjuan;Liu Long(Department of Health Statistics,School of Public Health,Shanxi Medical University(030001),Taiyuan)

机构地区：[1]山西医科大学公共卫生学院卫生统计学教研室,030001 [2]山西医科大学基础医学院数学教研室 [3]山西医科大学公共卫生学院儿少卫生学教研室 [4]重大疾病风险评估山西省重点实验室

出　　处：《中国卫生统计》2024年第2期227-230,234,共5页Chinese Journal of Health Statistics

基　　金：国家自然科学基金面上项目(81973154);山西省应用基础研究计划青年项目(201801D221399)。

摘　　要：目的将多状态Markov模型应用于阿尔茨海默病(Alzheimer's disease,AD)自然史研究,为慢性病自然史研究提供方法学借鉴。方法基于美国公共数据库ADNI,构建正常认知(normal cognition,NC)-轻度认知障碍(mild cognitive impairment,MCI)-AD多状态Markov模型,估计不同状态间的转移强度、转移概率、影响因素和各状态的逗留时间,绘制生存曲线,并进行模型拟合优度评价。结果性别、年龄、FAQ、MMSE、CDRSB、ADAS13、TMT-B-Time是AD自然史中重要的影响因素;NC与MCI状态的平均逗留时间分别为7.502年,11.621年。多状态Markov模型拟合结果较好。结论多状态Markov模型在AD等多状态慢性病转归研究中具有很好的应用价值。Objective The aim of this study was to explore the natural history of Alzheimer's disease(AD)based on the multi-state Markov model and to provide methodological reference for the chronic diseases.Methods Based on the American public database ADNI,we constructed a three states Markov model to estimate the transition intensity,transition probability,influencing factors and sojourn time of different states.Survival curves and goodness of fit was obtained.Results Gender,age,FAQ,MMSE,CDRSB,ADAS13 and TMT-B-Time were important influencing factors in the natural history of AD.The mean sojourn time of NC and MCI was 7.502 and 11.621 years,respectively.Model fit can be accepted.Conclusion Multi-state Markov model has a good application value for the outcome prediction of multi-state chronic diseases such as AD.

关 键 词：多状态MARKOV模型 阿尔茨海默病 慢性病 自然史 

分 类 号：R195.1[医药卫生—卫生统计学]