Generational trends in obesity in the United States: analysis with a wavelet coefficient self-organizing map

Increasing prevalence of obesity is considered to be a major public health problem, particularly in the United States [Flegel, C.M. et al., 2000 and Louise, W 2003], Many factors have been considered in causing this unhealthy condition, but an overall aging population could be a dominant cause, as it is well known that most people gain weight as they age. However, when the US population is analyzed as generational cohorts (based on year of birth), not simply as age cohorts growing older each year, some different patterns emerge. Trends based on age cohorts and generational cohorts are compared with geography (state) as a controlling factor. As trends in population weight can serve as a general health status "signal", a wavelet-based approach to analysis was selected, using Haar wavelet coefficients as the self-organizing map weight vectors. Similarity was determined by the Kullback-Leibler information statistic. The result is that generations prior to the "Baby Boomer" generation, who were not exposed to more recent unhealthy food consumption patterns as younger people, are less likely to be obese, in spite of their age. In addition, geography also plays a role. Understanding behavioral and attitudinal factors in obesity could lead to more targeted and effective public health campaigns.