Improved survival analyses in molecular cancer studies

Improved survival (whether this is literal survival, time to a change in symptoms, or time to disease progression) is a key outcome in cancer research. Most applications of survival analysis in the area use standard approaches, but the manner in which events are defined may bias these. Estimating and addressing such biases may be one strand of this project.

The context of these analyses is often one of using large (incorporating thousands to millions of molecular measurements, or different types, per patient) and potentially messy data sets. Other strands of this project might include the development of (asymptotically) efficient methods for survival analysis that make use of machine learning methods to exploit the amount of data present.

Alternatively, there may be investigations into survival-specific dimension reduction methods that can make such data sets more easily used and understood, perhaps incorporating prior information about the relationships between different levels of molecular data (e.g. DNA, RNA, protein) or prior knowledge of gene interaction networks.