Name: Farbod Nicholas Rahaghi
Email: frahaghi @ ucsd.edu
Grad Year: 2006
Understanding of normal blood glucose behavior is key to the setting of control criteria, modeling of disease onset and progress as well as designing more sensitive diagnostic criteria. Traditionally, assessment has been made by the study of HbA1c, which represents a surrogate measurement of the mean glucose values over the course of 6-8 weeks. Additionally, the glucose tolerance test, a dynamic test studying the time-series response of an individual to a glucose challenge has been used for diagnosis and monitoring purposes. With continuous glucose monitors on the horizon, the use of time-series analysis for the study of glycemic state is made possible. We obtained time-series of normoglycemic individuals from various sources including the literature and current ongoing studies. These were direct blood measurements and were in many cases were sampled very frequently (10-20mins). The time-scale of dynamics including daily and hourly patterns were studied in the different cases using various signal decomposition methods including Short-Time Fourier Transforms (STFT), Wavelet decomposition as well as Hilbert-Huang Transform. Statistical as well as dynamic parameters such as degree of nonlinearity, degree of stationary behavior, and higher order statistics were also analyzed. Dynamic parameters such as signal complexity, information content, and predictability were also assessed. A summary of findings defining the boundaries of normal glycemic behavior will be presented and possible utility in defining control criteria, diagnostic criteria as well as sensor criteria will be discussed.
<< Back to Posters or Search Results