Spatial Estimation of Radon Exposure for Epidemiologic Risk Assessment (2023)

Undergraduate: Kyle Sorensen

Faculty Advisor: Richard Smith
Department: Epidemiology (Gillings School of Public Health), Statistics and Operations Research

---

Spatial Estimation of Radon Exposure for Epidemiologic Risk Assessment:

Radon is a naturally occurring radioactive gas and is an intermediate product of the decay of uranium. Exposure to radon is the second leading cause of lung cancer in the United States and is hypothesized to cause strokes and other cardiovascular events. Additionally, radon levels seem to be rising across North America and may be linked to climate change. In the early 1990’s, the US EPA created a map of three distinct radon zones, classified according to indoor radon measurements (pCi/L) from the State Residential Radon Survey (SRRS), aerial radioactivity (ppm eU), geology, soil permeability and architecture type called the geologic radon potential (GRP) map. The goal of this analysis is to create an improved, granular spatial model for the geographic distribution of radon based on the SRRS data while accounting for other spatially dependent factors included in the calculation of GRP. The models used in this study include kriging, latent process modeling, LOESS and an ensemble estimation approach. The best model among these in terms of predictive accuracy was the ensemble estimation approach, with a mean absolute error of 2.059 pCi/L when fit to a 3 by 3 coordinate region in middle Tennessee. Future work on this project may include integration of additional data sets, inclusion of a temporal component leveraging more recent radon measurement data, further development of the ensemble estimation approach, accounting for the bias in the sampling design of the SRRS, or incorporating additional modeling approaches such as inverse distance weighted mean and nearest neighboring measure.

Link to Poster