Deb Sinha

Department of Biometry, Medical University of South Carolina


"Bayesian Analysis of Survival Data with a Surviving Fraction"

Survival data with a cured fraction commonly arises in cancer and other clinical trials, in which patients are considered `cured" after sufficient follow-up and in reliability studies when a machine may not fail during its technological lifetime. For modeling such data, we consider a new latent variable method which has useful properties for Bayesian purposes. We also compare our model with currently popular mixture model. We present extensions of our model to multivariate setting and to accelerated testing. Our methodologies are illustrated through the analyses of some cancer clinical trials data and reliability experiment involving highly reliable circuits.


Back to Colloquium Series