Semiparametric multi-parameter regression survival modelling

TL;DR

This paper introduces a flexible semiparametric survival model with covariate-dependent parameters, develops estimation methods with asymptotic properties, and demonstrates its effectiveness through simulations and real lung cancer data analysis.

Contribution

It proposes a novel semiparametric multi-parameter regression model for survival data with unspecified baseline hazard, including estimation procedures and variance estimation techniques.

Findings

Estimators have desirable asymptotic properties.

Simulation studies show good finite sample performance.

Application to lung cancer data demonstrates practical utility.

Abstract

We consider a log-linear model for survival data, where both the location and scale parameters depend on covariates and the baseline hazard function is completely unspecified. This model provides the flexibility needed to capture many interesting features of survival data at a relatively low cost in model complexity. Estimation procedures are developed and asymptotic properties of the resulting estimators are derived using empirical process theory. Finally, a resampling procedure is developed to estimate the limiting variances of the estimators. The finite sample properties of the estimators are investigated by way of a simulation study, and a practical application to lung cancer data is illustrated.