A0983
Title: Modeling past event feedback through biomarker dynamics in the multistate event analysis for cardiovascular disease data
Authors: Jianxin Pan - The University of Manchester (United Kingdom)
Chuoxin Ma - Beijing Normal University-Hong Kong Baptist University United International College (China) [presenting]
Abstract: In cardiovascular studies, ordered multiple events along disease progression are observed, which are essentially a series of recurrent events and terminal events with competing risk structures. One of the main interests is to explore the event-specific association with the dynamics of longitudinal biomarkers. A new statistical challenge arises when the biomarkers carry information from the past event history, providing feedback for the occurrences of future events and particularly when these biomarkers are only intermittently observed with measurement errors. A novel modelling framework isis proposed where the recurrent events and terminal events are modelled as multistate processes, and random effects models describe the longitudinal covariates that account for event feedback. Flexible models with semiparametric coefficients are adopted, considering the nature of long-term observation in cardiac studies. A one-step estimator of the regression coefficients is developed to improve computation efficiency, and their asymptotic variances for the computation of the confidence intervals are derived based on the proposed asymptotically unbiased estimating equation.