中文摘要：

目的将分段常数强度Markov模型应用于轻度认知损害（mild cognitive impairment,MCI）向阿尔茨海默病（Alzheimer＇s disease,AD）转归过程中,深入研究影响转归过程的因素,为制定不同发展阶段的预防措施提供理论依据,为其他多状态慢性病不同发展阶段影响因素的探讨提供方法学借鉴。方法应用太原市600名社区老年人的4次随访资料,以MCI为状态1,中重度认知损害为状态2,AD为状态3,拟合分段常数强度Markov模型,分析MCI向AD转归不同发展阶段的影响因素,并根据模型计算3年转移概率矩阵。结果经假设检验,数据满足Markov性（P=0.89）,不满足时齐性（P〈0.001）,应用分段常数强度Markov模型拟合,经多因素筛选,女性、年龄、吸烟、高血压和糖尿病是MCI向AD转归的危险因素,高文化程度和从事脑力劳动是MCI向AD转归的保护因素。由中重度认知损害向AD的转移概率随着随访时间的增加而增加。结论在数据满足Markov性,不满足时齐性时,分段常数强度Markov模型是对疾病转归过程的相关影响因素及其变化规律的有效分析方法。

英文摘要：

Objective The aim of this study was to introduce piecewise constant intensities Markov model in outcome prediction from mild cognitive impairment（ M CI） to Alzheimer＇s disease（ AD） and to find out related factors in order to provide theory basis for AD prevention among various progressive stages. The suggested method for exploring influencing factors for various progressive stages of other chronic disease w as also provided. Methods Our data came from four w aves of cohort study of600 community dw elling older people in Taiyuan. M CI,moderate / severe cognitive impairment,and AD w ere defined as state 1,2 and 3,respectively. Piecew ise constant intensities M arkov model w as applied to explore factors for various progressive stages from M CI to AD. According to the fitted model,three years transition probabilities among states w ere also estimated. Results Based on hypothesis testing,the M arkov assumption w as satisfied（ P = 0. 89） and the time-homogeneous assumption w as not（ P 0. 001）,so piecewise constant intensities Markov model was applied. Multivariate analysis showed that women,older,smoking,hypertension and diabetes w ere risk factors for progression from M CI to AD w hile high education and intellectual w ork w ere protecting factors. The transition probability from moderate / severe cognitive impairment to AD may increase as follow-up time extending. Conclusion Piecew ise constant intensities M arkov model is an effective analysis method to the data w hen the M arkov assumption w as satisfied and the time-homogeneous assumption w as not for related factors analysis and variation pattern during disease progressive process.