Authors: Elijah T. Johnson*, Auburn University, Nicholas Soltis, Auburn University, Stephanie R. Rogers, Auburn University, Ann S. Ojeda, Auburn University
Topics: Medical and Health Geography
Keywords: GIS, interpolation, geohealth
Session Type: Poster
Presentation File: No File Uploaded
Geographic information systems (GIS) has broad applicability in geohealth research to visualize and analyze the spatial distribution of diseases. However, publicly-available health data at certain census levels, like zip codes, can be incomplete. Agencies often do not report results for geographic regions with few datapoints, suppressing the data to preserve the confidentiality of the persons living there. At times, it may be impractical for health geographers to collect this information, so understanding ways to still use partial datasets and get highly accurate results is important. The purpose of this exploratory study was to compare two interpolation techniques, kriging and inverse distance weighting, to determine the best way to handle suppressed incidents of end-stage renal disease (ESRD) rates in Texas, Louisiana, and Arkansas at the zip code level. We also explored the variability of interpolation results using zip code data as polygons versus centroids. Additionally, we substituted certain values (null, 1, and 10) for the suppressed data points. Mean error and RMSE were generated and compared for each interpolation method. The results showed that kriging used with polygon zip code data and suppressed data points substituted with “1” had the lowest calculated error values and thus is considered to be the best technique for this study. The results of this exploratory study can be applied by researchers who prefer the most detailed datasets but may only have access to incomplete publicly-available health data.
To access contact information login