Online Program Home
My Program

Abstract Details

Activity Number: 426 - SPEED: Biopharmaceutical and General Health Studies: Statistical Methods and Applications, Part 2
Type: Contributed
Date/Time: Tuesday, July 30, 2019 : 3:05 PM to 3:50 PM
Sponsor: Health Policy Statistics Section
Abstract #307841
Title: Sieve Maximum Likelihood Method for Interval-Censored Data with Missing Covariates Under Proportional Hazards Model
Author(s): Ruiwen Zhou* and Huiqiong Li and (Tony) Jianguo Sun
Companies: University of Missouri-Columbia and Yunnan University and University of Missouri
Keywords: Missing at random; Interval-censored data; Proportional hazards model; EM algorithm; Monte Carlo EM algorithm

This project considers the regression analysis of interval-censored data when the covariates are missing at random (MAR). The proposed method used I-spline to approximate the unknown nondecreasing cumulative baseline hazard function. Formulate the PH model in this fashion results in a finite number of parameters to estimate while maintaining substantial modeling flexibility. (Wang, et al. 2016). To avoid the complexity of finding the maximum likelihood estimation of the parameters, we developed a weighted expectation-maximization (EM) algorithm to fi nd the MLE. The derivation of EM algorithm relies on two-stage data augmentation on latent Poisson random variables. For the missing covariates, the developed method allows us to handle both categorical covariates and continuous covariates. When the missing covariates are continuous, we adapt Monte Carlo version of EM algorithm to estimate the parameters.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program