A0640
Title: Semiparametric joint modeling for biomarker trajectory before disease onset
Authors: Yifei Sun - Columbia University (United States) [presenting]
Abstract: Understanding the dynamics of biomarkers prior to disease onset is a critical topic in biomedical research. A semiparametric joint model is proposed to analyze the temporal evolution of biomarkers, allowing for a flexible biomarker trajectory shape that depends on two-time scales: a natural time scale such as age and the time relative to disease onset. An additional complication arises because the natural time scale often differs from the time of the study, leading to analytical challenges such as left-truncation bias and irregular measurements. To address these issues, a profile kernel estimating equation approach is introduced to estimate regression coefficients and unspecified baseline mean trajectory functions. The large-sample properties are established of the proposed estimators and conduct simulation studies to evaluate their finite sample performances. The method is applied to investigate the brain biomarker trajectory before the onset of preclinical Alzheimer's disease.
