Colloquium: Wei-wen Hsu

Speaker: Dr. Wei-wen Hsu, Kansas State University

Title: A Sup-Score Test for the Cure Fraction in Mixture Models for Long-Term Survivors 

Abstract: Score tests are often used to assess the cure proportion in the population under interesting two-component mixture models for censored survival data with long term follow-up. But the existing test procedures often rely on restrictive assumptions, particularly on the form of the alternative. One common restriction is that constant cure proportions are often assumed. In this paper, we adopt a general formulation where these proportions are allowed to depend on covariates via a regression model. The hypothesis then translates into testing infinitely large intercepts in the mixing proportions while holding fixed the other regression terms. The implied hypotheses are not typical and standard regularity conditions to conduct the test may not even hold. Using empirical processes arguments, we construct a score-based test statistic under this general formulation and establish its limiting null distribution as a functional of chi-square processes. In practice, we suggest a simple resampling procedure to approximate this limiting null distribution. Our simulation results show that the proposed test can greatly improve efficiency over tests based on constant cure proportions under the alternative. The practical utility of the methodology is illustrated using ovarian cancer survival data with long-term follow-up from the SEER registry.

Faculty host: Guanqun Cao

time (1 pm) and place (224), and refreshment time (2pm)