Data Mining of Hierarchical Determinants of the Obesity Epidemic
College: Applied Health Sciences
Award year: 2001-2002
The prevalence of obesity in the United States continued to grow during 1990s, representing a serious public health threat to millions of Americans. Many factors, or determinants, are believed to contribute to the current obesity epidemic and they are nested in a hierarchical structure (e.g., individuals' exercise and eating behaviors are nested in their family and friends' social support systems, which are further nested in communities with/without environmental barriers). This hierarchical relationship, however, has not been fully recognized and quantitatively determined, due to perhaps the complexity of the relationship and analytical difficulty. Using data mining, a new computer-based analytical method, the purpose of this collaborative project is to examine the hierarchical determinants of the obesity epidemic by connecting and data mining several major national databases, such as the National Health and Nutrition Examination Survey, the Behavioral Risk Factor Surveillance System, and the National Personal Transportation Survey. The major tasks of this project include: (a) identify and become familiar with related national databases that may contain information on potential determinants of obesity epidemic; (b) conduct data mining of each database individually to fully understand the variables and possible determinants, as well as their interrelationships, within the database; and (c) link identified databases together and determine hierarchical relationships among determinants cross databases. The data analysis will be completed using data mining and other quantitative analytical techniques, e.g., the hierarchical linear model. Upon the completion of this project, a much better understanding of these hierarchical determinants is expected.