對主要終點為設限存活時間的三臂非劣臨床試驗設計使用參數與半母數模型

In this project, we will investigate the gold standard design for three-arm noninferiority trials, which includes an experimental group, a reference group and a placebo group using parametric and semiparametric models. Four parametric models (lognormal, Gompertz, Rayleigh, and loglogistic) and three semiparametric models (Cox, accelerated failure time and transformation models) will be considered for the primary survival time endpoint. Minimum required sample size and testing power with optimal allocation of the three groups will be considered as the criteria of performance of both parametric and semiparametric method. In particular, we will investigate the impact of the situations under delayed treatment effects and model misspecification. Simulation studies will be conducted to find out the performance of all methods.