Local likelihood in regression analysis of proportional mean residual life model with censored survival data

As a function of time t, mean residual life (MRL) is the remaining life expectancy of a subject given its survival to t. In survival analysis, the relationship between a survival time and a covariate can be conveniently modeled with the proportionality mean residual life (MRL) model proposed by Oakes and Dasu (1990) and provides an alternative to the Cox proportionality hazards model, Cox (1972). In this paper we consider the proportional MRL regression model with a nonparametric covariate effect. We discuss estimation of the proportional function when the baseline MRL function is not specified. We develop the asymptotic properties of the proportionality function. Simulation studies are presented to assess the finite sample behavior.