Modeling past event feedback through biomarker dynamics in the multistate event analysis for cardiovascular disease data

In cardiovascular studies we often observe ordered multiple events along disease progression which are, essentially, a series of recurrent and terminal with competing risk structure. One the main interests is to explore event specific association dynamics longitudinal biomarkers. A new statistical challenge arises when biomarkers carry information from past history, providing feedbacks for occurrences future and, particularly, these are only intermittently observed measurement errors. this paper propose novel modeling framework where modeled as multistate processes covariates that account described by random effects models. Considering nature long-term observation in cardiac studies, flexible models semiparametric coefficients adopted. To improve computation efficiency, develop an one-step estimator regression derive their asymptotic variances confidence intervals, based on proposed asymptotically unbiased estimating equation. Simulation show naive estimators, either ignore or errors, biased. Our method achieves better coverage probability, compared methods. The model motivated applied dataset Atherosclerosis Risk Communities Study.