TITLE:
A Simulation Study on Comparing General Class of Semiparametric Transformation Models for Survival Outcome with Time-Varying Coefficients and Covariates
AUTHORS:
Yemane Hailu Fissuh, Tsegay Giday Woldu, Idriss Abdelmajid Idriss Ahmed, Abebe Zewdie Kebebe
KEYWORDS:
Estimating Equation, Semiparametric Transformation Models, Time-to-Event Outcomes, Time-Varying Coefficients, Time-Varying Covariate
JOURNAL NAME:
Open Journal of Statistics,
Vol.9 No.2,
April
2,
2019
ABSTRACT: The consideration of the time-varying covariate and time-varying
coefficient effect in survival models are plausible and robust techniques. Such
kind of analysis can be carried out with a general class of semiparametric
transformation models. The aim of this article is to develop modified
estimating equations under semiparametric transformation models of survival
time with time-varying coefficient effect and time-varying continuous
covariates. For this, it is important to organize the data in a counting process
style and transform the time with standard transformation classes which shall
be applied in this article. In the situation when the effect of coefficient and
covariates change over time, the widely used maximum likelihood estimation
method becomes more complex and burdensome in estimating consistent estimates.
To overcome this problem, alternatively, the modified estimating equations were
applied to estimate the unknown parameters and unspecified monotone
transformation functions. The estimating equations were modified to incorporate
the time-varying effect in both coefficient and covariates. The performance of
the proposed methods is tested through a simulation study. To sum up the study,
the effect of possibly time-varying covariates and time-varying coefficients
was evaluated in some special cases of semiparametric transformation models.
Finally, the results have shown that the role of the time-varying covariate in
the semiparametric transformation models was plausible and credible.