-----------------------------------------------------------

Statistics Seminar Series

Thursday, November 16, 2006 @ 4:00PM
Cullimore Hall, Room 611
New Jersey Institute of Technology

A Statistical Model for Inter-temporal Causal Dynamics of Glucose Production and Disposal with Insulin Level in Diabetic Patients

(Accepted for publication in Statistics in Medicine, Special Issue)

  Das Purkayastha, Ph.D.
Associate Director, Biostatistics

¹ Biometrics, U.S. Clinical Development and Medical Affairs, Novartis Pharmaceuticals
One Health Plaza, East Hanover, NJ 07936

This paper provides a useful tool to examine inter-temporal behavior in glucose and insulin level in diabetic patients. The primary thrust of this paper is to model inter-temporal causal dynamics of endogenous and ingested glucose production and disposal with insulin in diabetic patients. A short-run compartmental dynamics of glucose and insulin was introduced with their stochastic relationship through a vector autoregressive (VAR) scheme. This study also aims to develop glucose variability indices and test for their independence, homogeneity and symmetry at steady state condition. The model also empirically identifies the significant prognostic factors and other exogenous variables affecting the dynamics. An empirical application provides more compelling evidence of such dynamics. The estimated model reconfirms the fact that pre-prandial PM glucose level has statistically significant effects on post-prandial PM glucose level in diabetic patients. Also pre-prandial PM insulin level shows a significant impact on post-prandial PM insulin in this study. In addition, results also indicate statistical significance of body mass index (BMI) and hemoglobin on post-prandial PM glucose level.  It is interesting to note that the pattern of glucose variability indices in pre-and post- prandial AM time points are different from those at PM period.  The results also show interdependence between the glucose variability indices.  Only the pre- and post-meal PM variations of glucose are homogenous to its pre- and post-meal AM variations.

KEY WORDS:  pre- and post-prandial glucose; vector autoregressive model; simultaneous equation; glucose variability indices; inter-temporal dynamics; stochastic specifications;