A0645
Title: Survival analysis with a random change-point
Authors: Chun Yin Lee - Hong Kong Polytechnic University (Hong Kong) [presenting]
Kin Yau Wong - Hong Kong Polytechnic University (Hong Kong)
Abstract: Contemporary works in change-point survival models mainly focus on an unknown universal change-point shared by the whole study population. However, in some situations, the change-point is plausibly individual-specific, such as when it corresponds to the telomere length or menopausal age. Also, maximum-likelihood-based inference for the fixed change-point parameter is notoriously complicated. The asymptotic distribution of the maximum likelihood estimator is non-standard, and computationally intensive bootstrap techniques are commonly used to retrieve its sampling distribution. The motivation is from a breast cancer study, where the disease-free survival time of the patients is postulated to be regulated by the menopausal age, which is unobserved. As menopausal age varies across patients, a fixed change-point survival model may be inadequate. Therefore, a novel proportional hazards model is proposed with a random change-point. A nonparametric maximum likelihood estimation approach is developed, and a stable EM algorithm is devised to compute the estimators. Because the model is regular, conventional likelihood theory is employed for inference based on the asymptotic normality of the Euclidean parameter estimates, and a profile-likelihood approach can consistently estimate the variance of the asymptotic distribution. A simulation study demonstrates the proposed methods' satisfactory finite-sample performance, yielding small bias and proper coverage probabilities.
