High-dimensional marginal screening for right-censored survival data

This talk discusses a marginal screening test to detect the presence of
significant predictors for a right-censored time-to-event outcome under a
high-dimensional accelerated failure time model. Establishing a rigorous
screening test in this setting is challenging, not only because of the right
censoring, but also due to the post-selection inference in the sense that an
implicit variable selection step needs to be taken into account to avoid
inflating the Type I error.

We develop two approaches:

1) a maximally-selected inverse-probability-of-censoring weighted test
statistic based on marginal correlation, and

2) a regularized semi-parametrically efficient augmentation of 1) that is robust
to misspecification of the model used for the censoring distribution.

The pros and cons of each of these approaches are discussed.

The talk is based on joint work with Tzu-Jung Huang, Alex Luedtke and Min Qian.