Abstract:
|
Cancer epidemiology studies often examine multiple genetic and exposure factors in relation to disease outcome. In several studies. exposures are selected according to certain themes of interest. Higher level biological information about genetic factors are also increasingly available from the public domain or previous studies. Incorporating these themes into the analyses constitutes a useful data reduction technique. The resulting analysis, referred to as reduced rank regression approach, can lead to more precise estimates of risk parameters than standard methods that ignore these themes. This talk presents different estimation techniques for analyzing data in such settings, evaluates the operating characteristics of these methods, and provides empirical illustration using data from published epidemiology studies of nevi and cognitive functions.
|