Statistical Methods for Medical Studies

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The aim of this competing renewal is the development of statistical methods for biomedical research, with principal focus on techniques and tools needed for chronic disease population research. There is a major emphasis on statistical design and analysis procedures for using the high-dimensional genomic and proteomic data that are coming available. These data have much potential to stimulate various important research areas including early detection of disease, disease risk profiling, preventive intervention development, and the elucidation of preventive intervention effects. There is also a continuing emphasis on the modeling and use of data on important environmental (e.g., dietary) exposures; on the avoidance of bias under several important study designs; and on the interplay among such designs in the chronic disease population research agenda. The Program will continue the current three projects and administrative core. Project 1 is concerned with statistical topics relevant to epidemiologic cohort studies and disease prevention trials. These include several topics in the analysis of failure time data; methods for dietary and physical activity measurement error accommodation; design and analysis methods for marginal effects in genome-wide single nucleotide polymorphism association studies (GWAS); and study of differential biases between cohort studies and randomized controlled trials. Project 2 focuses primarily on analyses beyond marginal associations for GWAS. Topics include combination of data from population-based and family-based association studies; the identification and assessment of gene-gene and gene-environment interactions: and the simultaneous estimation of linkage and haplotype associations in multipoint analysis of affected sib-pairs. Project 3 is concerned with biomarker discovery and evaluation methods for early detection of disease and for other purposes. Topics include design and analysis methods to distinguish cases from controls based on functional data (e.g., mass spectra) as arise, for example, in proteomic research; the development of methods to assess the predictive value of disease biomarkers: and the study of sequential designs for biomarker selection and validation. Collectively, these projects will apply the talents of 14 committed and interactive statistical and mathematical scientists to address statistical topics that are among the most important for progress in chronic disease population research.