Statistical Methods in Cancer Research

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): A team of biostatistical methodologists propose to develop new methodology for the statistical analysis of cancer data and cancer studies. The long-term objective is to impact our understanding of familial associations, diagnosis, and interpretation of survival data in the context of cancer studies. In particular, the investigators propose to establish semiparametric variance component models as a robust data analytic technique in haplotype analysis of disease-related familial traits. In regard to cancer survival data, they intend to develop a general class of hazard models when the data do not exhibit proportional hazards or constant relative risk. In cancer diagnosis, they propose to develop novel nonparametric techniques to analyze longitudinal receiver operating characteristic curves (ROC). Finally they intend to impact the design of oncology trials by developing efficient allocation ratios and a sequential monitoring plan for diagnostic trials comparing ROC curves. Methods will be tested and illustrated using oncology data from several sources, including extensive cancer biomarker data available directly at George Mason University. PUBLIC HEALTH RELEVANCE: A team of biostatisticians proposes to develop new methodology to analyze data arising from cancer studies. The proposed methods will facilitate analysis of survival rates in cancer, identification of familial associations, and aid in clinical trials investigating diagnostic methods in cancer detection.