A1220
Title: On a new EM algorithm for semi-parametric Cox models under interval censoring
Authors: Jun Ma - Macquarie University (Australia) [presenting]
Abstract: Interval-censored event times are common, particularly in studies of chronic diseases. Currently, estimation methods for semi-parametric Cox models with partly interval-censored event times include EM algorithms and direct-constrained optimization techniques. A new EM approach is introduced that treats event times for left- and interval-censored data as missing values. In each iteration, the E-step involves computing the conditional expectation of the complete-data log-likelihood, which has a closed-form expression. In the M-step, a profile likelihood is used to efficiently update the regression coefficients and the baseline hazard function, which is approximated using a piecewise constant function. This iterative process typically converges quickly. A simulation study and a real-data example are also presented to illustrate the performance of the proposed method.
