This project focuses on the application of joint modeling techniques to integrate longitudinal and survival data, with a specific emphasis on clinical research. The methodology involves specifying two interconnected sub-models: a longitudinal model to capture the trajectory of longitudinal measurements over time and a time-to-event model to describe the hazard or survival function. The longitudinal sub-models may include linear mixed-effects models, generalized linear mixed models, or other suitable techniques, while the time-to-event sub-models commonly employ Cox proportional hazards models, parametric survival models, or flexible parametric models. The association between the longitudinal and event processes is accounted for by incorporating shared random effects, which capture individual-specific characteristics influencing both the longitudinal response and the event occurrence. Estimation of the joint model parameters is achieved through maximum likelihood estimation or Bayesian methods, such as Markov chain Monte Carlo (MCMC) techniques. Model assessment includes checking assumptions, evaluating model fit, and assessing predictive performance using techniques like residual analysis, deviance information criterion (DIC), and simulation studies. The project also involves comparing the performance of the joint model with alternative models to assess the added value of the joint modeling framework. The ultimate goal is to enhance the understanding of the interrelationship between longitudinal and survival outcomes in clinical research, providing more accurate and insightful predictions of patient trajectories and survival probabilities.

Dr. Mini Jayan (Guide)