Abstract
Objectives
In chronic diseases with metabolic/nutritional disturbances, longitudinal analysis of the pattern of disease development prior to reaching the overt diagnosis, is challenging. In the context of studying the effects of food intake on the longitudinal development of obesity, and type 2 diabetes, we developed and tested a new method of analysis. These disorders have significant inter- and intra-individual differences in 1) the age of onset 2) the rate of progression from initial signal to diagnosis; 3) the timing of significant changes in this progression; and 4) the varying severity or amplitudes of indices of progression.
Methods
In order to make pattern recognition possible, we have first defined the critical diagnostic features (e.g., in the case of diabetes, a fasting plasma glucose level ≥ 126 mg/dl (defined by us as phase 8); in the case of obesity, a BMI ≥ 27) as the critical data point, and subsequently we identified all distinct phases.
Results
We have concluded that some variables were critical, and measurable, and had to be weighted differentially in the disease development (e.g., Phase 1 age < 10yrs age, but thereafter uninformative), while others emerged in midcourse (phases 3, 4, and 5, e.g., beta cell function), or immediately proximal to disease diagnosis (phases 6 and 7, e.g., acute insulin response loss). This new phase approach to pattern recognition permits improved characterization of a disease prodrome, but also allows wide variations in age of onset and rates of progression, and in the amplitude of the changes within features. Across these phases, variables have different impacts and are weighted differently. Nevertheless, the sequence of appearance of these features was invariable, much like an accordion, and this new method has allowed for early pattern recognition during this pre-disease period.
Conclusions
This analytical method is essential to recognizing the earliest features in disease development, and therefore the potential for early effective targeting to slow progression. Variables are simultaneously examined in the context of their sequence of expression, with unexpected potentially impactful variables identified, and potential causes and effects noted.
Funding Sources
National Institute on Aging HHSN263200800022C and the University of South Florida Foundation.