Time Transformation Intensities Markov Model In Outcome Of Mild Cognitive Impairment To Alzheimer’s Disease

Objective:Based on a time transformation Markov model, the aims of this study were to find out related factors which influence the risk of progression for various progressive stages and to make prediction in outcome of mild cognitive impairment(MCI) to Alzheimer’s disease(AD). Methods:Our data came from five waves of cohort study of 518 community elders in Taiyuan by the end of 2012. The progression from MCI to AD were defined as MCI(state 1), moderate to severe cognitive impairment(state 2), and AD(state 3). A three state time transformation Markov model was applied to explore influencing factors at different disease stages and to make outcome prediction.Results:Based on hypothesis testing, the time-homogeneous assumption was not(P<0.005) and the Markov assumption was satisfied(P=0.468),so time transformation Markov model was applied. Multivariate analysis showed that gender, age, education, occupation, smoking and hypertension were statistically significant for transition from state 1(MCI) to state 2(moderate to severe cognitive impairment); Age, education, occupation, diabetes and hypertension were statistically significant for transition from state 1(MCI) to state 3(AD); Age, education, smoking diabetes and hypertension were statistically significant for transition from state 2(moderate to severe cognitive impairment) to state 3(AD). Therefore, women, older, smoking, diabetes, hypertension and low educational level were risk factors for progression from MCI to AD; while intellectual work was protecting factor for progression from MCI to AD. According to the fitted time transformation Markov model, survival curves, transition intensities, and three years transition probabilities among each state were also estimated. Conclusions:To delay the disease progression of AD, prevention measures phase by phase can be taken based on the main factors of each stage. Compared with homogeneous Markov model, the time transformation Markov model is more all-around in the related factors analysis and variation pattern during disease progressive process. The appropriate model should be selected based on the collected data.