Multi-state Markov Model In Outcome Of Mild Cognitive Impairment

Objective:The aim of this study is to introduce Multi-state Markov Model in outcome prediction of mild cognitive impairment and in-depth analyze the longitudinal changes of MCI patients'cognitive level and related factors in order to provide theory basis for AD prevention and early intervention among elderly people. Also this study is designed to discuss about Multi-state Markov model methodologically and to provide methodological reference for the biology model research in which states can be used to express dynamic life processes.Methods:We supposed that during the progression of AD, transition from their present state to the next state only depends on the current state, but not directly affected by the previous states. This is called Markov "unfollow-up effect" property. And based on IQ changes that reflect trends in cognitive function in the follow-up patients, we constructed a four states model. According to the multi-state Markov model theory, parameters estimation method was determined, model parameters were obtained, and finally conclusions were drawn.Results:Multi-state Markov Model was fitted well. Univariate analysis showed that gender, age, education level, marital status, smoking, cerebral hemorrhage, hypertension, high cholesterol in the serum, diabetes, LDL, SBP, and DBP were statistically significant. For multivariate analysis, gender, age, education level, cerebral hemorrhage and SBP affected transition from state of cognitive function stability to state of severe deterioration. MCI patients of female, the older, lower educational level, the occurrence of brain hemorrhage, with hypertension were associated with an increasing risk of transition from state of cognitive function stability to state of severe deterioration. Age and education level influenced transition from state of cognitive function stability to state of slight deterioration. Education level influenced transition from state of cognitive function stability to state of improving. None of gender, age, education level, cerebral hemorrhage and SBP affected transitions between other states. And according to the fitted Multi-state Markov model, we estimated transition intensity, two years transition probability, and mean sojourn times during each state.Conclusion:Based on the main factors of each stage during progression of AD, prevention measures phases by phases can be taken. In addition, as an effective tool for dealing with longitudinal data, the multi-state Markov model can take into account all of the states, outcomes, time information of state transition and possible factors simultaneously. It can also find disease progression related factors, evaluate disease process dynamically, analyze transition probabilities between states and how these transition probabilities change with factors and time.