Sieve Estimation for the Proportional-Odds Failure-Time Regression Model with Interval Censoring

Abstract Estimation of the proportional-odds failure-time regression model with interval censoring is considered. Conditions that allow for positive information for the regression parameter are discussed. The efficient score is characterized by a Fredholm equation of the second kind. The sieve maximum likelihood estimator for the finite-dimensional regression parameter is shown to be asymptotically normal with √n convergence rate and to achieve the information bound. Data analysis and simulations assist in clarifying our thoughts regarding the choice of sieve for finite-sample problems.