Semiparametric model of mean residual life with biased sampling data

The mean residual life function is an important and attractive alternative to the survival function or the hazard function of survival time in practice. It describes the remaining life expectancy of a subject surviving up to time t. To study the relationship between the mean residual life and its associated covariates, a broad class of mean residual life models under general biased sampling data is proposed in this paper, thereby extending the reach of this flexible and powerful tool. The unknown parameters are estimated by using inverse probability weighting method. An easily used model diagnostic method is also presented to assess the adequacy of the model. Both asymptotic properties and finite sample performances of the proposed estimators are established. Finally the practical appeal of the estimator is shown via two real applications using the Channing House data and the Canadian Study of Health and Aging (CSHA) dementia data.