Short- and Long-Term Effects in Prostate Cancer Survival: Analysis of Treatment Efficacy and Risk Prediction

摘要

This project represents a successful effort to develop abstract statistical theory, computational algorithms, translate this methodology into stable shareware software products that can be used by the broad scientific community, put this product into the R-Projects nltm and rpNLTM that has become the dominant site for dissemination of cutting edge statistical procedures, and finally use and showcase all this arsenal to address real data and problems in prostate cancer. We are glad that we took the challenge of this large idea development and translational effort in the three year project performance period and that were able to see it through. The goal of this proposal was to investigate a novel approach to the analysis of post-treatment survival of prostate cancer patients: the decomposition of the diversity of survival patterns into short-term and long-term effects. We proposed to identify a model of prostate cancer survival incorporating long- and short-term effects of prognostic factors and treatment. Novel statistical tools were developed to make such models work for better prognosis of prostate cancer patients. Year 1 at the University of Utah was primarily devoted to development of methodology for point estimation and hypothesis testing. While continuing methodological research in Year 2, we focused on the delivery aspect of the project addressing software development and implementation of the algorithms, testing them by simulations, development of tools for multivariate analysis and variable selection and preliminary applications of these tools to real data.