Kristen Hampton presents MSPH Technical Report Final Defense

December 09, 2005
Kristen Hampton presents her MSPH Technical Report Final Defense on Monday, December 12th at 1:00PM in 2304 McGavran-Greenberg. Full details follow.Adjusting for sampling variability in sparse data: A Bayesian Maximum Entropy approach to disease mapping
Under the direction of Marc L. Serre)

Disease maps summarizing the spatial and spatio-temporal variation in disease risk have a wide range of applications, from hypothesis generation to public health surveillance. The utility of disease mapping, however, depends upon how accurately the value being mapped estimates the spatial process of interest. Especially with rare diseases, calculating crude rates from routine health data indexed at a small geographical resolution poses specific statistical problems due to the sparse nature of the data, particularly observational noise due to sampling variability. Here we introduce a mathematically rigorous framework for quantifying the error in observed rates due to sampling variability and present a Bayesian Maximum Entropy (BME) approach to disease mapping in order to correct for this error. We demonstrate the effectiveness of the BME adjustment through in-depth analyses of both simulated data and real surveillance data of HIV disease in North Carolina. The aim is to produce more reliable maps of disease risk in small areas in order to improve identification of spatial and temporal trends at the local level.

Committee: Dr. Marc Serre, Advisor Dr. Dionne Law (NIEHS) Dr. William Miller (EPID)

For further information please contact Rebecca Riggsbee Lloyd by email at Rebecca_Lloyd@unc.edu