Dietary Patterns Analysis
Just as critical to existence as the air we breathe, is the food we eat. The physical structure that defines each one of us is made from the foods we have eaten and as such, food is a natural component of health and wellbeing. Day-to-day dietary choices affect current and future health. It has long been recognized that diet plays a direct role in 4 of the 10 leading causes of death in the United States. Chronic diseases such as heart disease, diabetes, and obesity are known to develop over many years and have been strongly linked with diet. These staggering statistics point to diet as a major player in health and having a pivotal role with regard to other behaviors, e.g., physical activity, social and economic issues. Thus, this proposal examines diet as the center for developing models for public health patterns and trends, as diet represents the "gorilla in the room." Epidemiology, the scientific method for the study of the distribution and determinants of disease frequency in human populations and translation to control health problems, has hit a wall with being able to document the true relationships between diet, other behaviors, and disease. The areas of statistical signal processing and machine learning hold promise to complement existing epidemiological approaches with the goal of providing realistic profiles of human behavior (with diet as the core) that can translate to evidence-based information that inspires the participation of the public in their healthcare and healthful lifestyles.
These diet-chronic disease relationships hint at "patterns" of dietary intake that differ from the dietary patterns of healthy individuals. Epidemiological research would suggest that dietary patterns are likely to play the dual role of protector and promoter in the development of disease. However, traditional analysis used in nutrition epidemiology may be an inefficient approach to address the multifaceted nature of diet-disease relationships. Factors involved in these processes include time and frequency of eating, the combinations of nutrients and foods eaten routinely, diet at certain critical life stages, timing and amount of exercise and sleep with regard to eating, and other personalized lifestyle habits such as where food is consumed (e.g. home or work) and work habits (e.g. shift workers). These individualized and interacting variables are broadly termed "dietary patterns." There is no standard definition in nutrition epidemiology of a "dietary pattern." For the purposes of our work we shall define a dietary pattern to include dietary evaluation in which multiple dietary characteristics (e.g. foods or nutrients) are examined simultaneously rather than individually.