Abstract:
To model time-to-event data, the Weibull, log-logistic, and log-normal distributions
are commonly utilized. Only monotone hazard rates are accommodated by the
Weibull family, although log-logistic and log-normal are extensively employed to
describe unimodal hazard functions. We need more flexible models that can in-
corporate both monotone and non-monotone hazard functions since lifespan data
with a wide variety of features is becoming more widely available. The extended
Exponential-Weibull distribution model, for example, not only supports monotone
hazard functions but also allows for bathtub and unimodal shape hazard rates. In
univariate study of time-to-event data, this distribution has shown a lot of promise.
Many research, on the other hand, are primarily concerned with determining the
link between the time it takes for an event to occur and one or more covariates.
In time-to-event analysis, this leads to the examination of survival regression mod-
els, which may be expressed in a variety of ways. Formulating models for the
accelerated failure time family of continuous distributions is one such method. The
Weibull, log-logistic, and log-normal distributions are the most widely utilized for
this purpose. In this paper, we show that the extended exponential-Weibull dis-
tribution is closed under the accelerated failure time framework. Then, using the
maximum likelihood approach, we build a survival regression model based on the
extended Exponential-Weibull distribution and estimate the model parameters. The
performance of the model parameter estimators was demonstrated using a compre-
hensive Monte Carlo simulation analysis. To demonstrate the applicability of the
novel proposed survival regression model, two real-life survival data sets from cancer therapies were used. The simulation and real-world applications show that the
proposed model is capable of accurately characterizing various forms of time-to-
event data.
