Post date: Mar 19, 2015 2:37:27 PM
Title: Nonparametric Methods for Interval Censored Data
Speaker: Michael Fay, Biostatistics Research Branch, National Institute of Allergy and Infectious Diseases, NIH.
Interval censored data are time to event data where for each observation we do not observe the event but only know that it is bracketed between two assessment times. For example, if the event is time until HIV is first detectable in the blood, then we only know that the event happened after the last blood draw with no detectable HIV and before the first blood draw with detectable HIV. The talk will be a review, focusing on the nonparametric maximum likelihood estimator of survival and different types of weighted logrank tests. I will discuss the validity of testing treatment effects using different logrank implementations (e.g., permutation tests, score tests, multiple imputation tests) when assessment times may depend on treatment.