Aelita Huang
PhD Candidate, University of Waterloo
Room: M3 3127
Modeling and Testing Effects of HLA Markers on Damage Processes in Individual Joints under Intermittent Observatio
Psoriatic arthritis is associated with increased risk of damage in hand, foot, and sacroiliac joints. Progression occurs across multiple anatomical sites and features clustering within patients. These features complicate the evaluation of genetic markers, especially when marker effects may differ by joint type. We develop multistate models for multiple progressive processes observed intermittently over time and construct score tests for screening markers under several assumptions, including a) a common marker effect across joint types, b) joint-type-specific effects, and c) heterogeneity across joint types. Robust methods of estimation and inference are developed using generalized estimating equations for clustered Markov processes. Simulation studies show that estimates have negligible finite sample bias, there is close agreement between empirical and robust standard errors, and near-nominal coverage of 95% confidence intervals. For pseudo-score tests based on this formulation, the type I error rate is controlled, and good power is shown across a range of settings. The proposed methods provide more detailed joint-type-specific inference and more stable, interpretable results than analyses based on Cox models applied to patient level measures of joint progression.