Shared Frailty Model for Joint Survival Data - A Simulation Study

Performing a survival analysis, allows investigating factors that contribute to outcome over time. Though considerable amount of literature has addressed this area, analysis multiple responses not received sufficient attention in past literature. The joint modeling fits well when there are for the same study unit and it can provide improved results than fitting univariate models separately since correlation between be captured through model. Therefore, aim was propose approach, which linkage two responses(for simplicity, research bivariate lifetime data have been considered) derived by sharing common random effect under different distributions parametric forms baseline hazard function. In study, Gamma Normal Exponential Weibull were considered. performance Shared Frailty model compared with Ordinary Proportional Hazard simulation simulated three sample sizes 20% 40% censoring proportions structures. Bias, Coverage Probability (CP) Mean Squared Error (MSE) measures used. Parameter estimates showing relatively low bias, high CP, minimal MSE confirmed suitability proposed capture data, surpassing fit models. Also, interpreted, distributed effects more suitable times while times.