Advanced Joint Survival Models for Evaluating Treatment Efficacy

Abstract

This study proposes a joint modelling framework for evaluating treatment efficacy in AIDS by integrating longitudinal and survival data. The approach addresses a key limitation in biomedical analysis by incorporating left-censored covariates into a unified likelihood function, ensuring more accurate and unbiased parameter estimation. A linear mixed-effects model is used to describe longitudinal biomarker dynamics, while a Weibull frailty model captures time-to-event outcomes and unobserved heterogeneity. The performance of the proposed model is assessed through Monte Carlo simulations under varying sample sizes and hazard conditions. Results indicate improved estimation accuracy, reduced bias, and greater robustness compared to conventional methods, particularly in the presence of censored or incomplete data. In addition, the framework effectively captures the association between longitudinal processes and survival outcomes, offering deeper insight into disease progression. Overall, the proposed method provides a reliable and flexible tool for clinical data analysis. It supports more informed decision-making in treatment evaluation and has potential applications in AIDS research and other chronic disease studies.