Kenta Yoshida, PhD: No relevant disclosure to display
Statistics and Pharmacometrics (SxP) SIG Presentation Parametric survival models, such as the tumor growth inhibition-overall survival (TGI-OS) model, is an important tool for performing clinical trial simulation to guide clinical drug development decisions. When performing clinical trial simulation, incorporation of uncertainty in parameter estimates is a necessary step to translate the level of confidence from the data used in model development into the precision of the simulated outcomes. We developed an R package survParamSim to facilitate the simulation from parametric survival models (accelerated failure time models in survreg() function of survival package). Importantly, parameter values are resampled from the variance-covariance matrix of the parameter estimates to incorporate parameter uncertainty when performing repeated simulations. The package also implements useful post-processing tools, such as derivation of Kaplan-Meier curves, hazard ratios from the nonparametric proportional hazard model, and their prediction intervals from repeated simulations, as well as clinical trial simulation with resampled patient populations. This package has greatly enhanced the workflow of performing TGI-OS modeling, contributing to wider application of this approach.
