Scalable computation for geospatial exposure assessment with R and high-performance computing
This project aims to develop a R package for geospatial exposure assessment for climate and health studies by enabling easy-to-use and readily parallelized functions with extensive datasets and high-performance computing. The output is expected to lower barriers in large-scale health effect analyses where leverage geospatial exposure / geomarkers.
Causal inference for the relationship between deaths by mental illness and residential greenspace exposure with matching methods by a combined matrix of Jensen-Shannon divergence and geographic distance
I am integrating the multivariate dissimilarity and the spatial contiguity to match the data better for causal inference of the relationship between deaths by mental illness and residential greenspace exposure while controlling confounding environmental exposures, individual traits, and neighborhood socioeconomic status in the U.S. Pacific Northwest.
Machine learning and spatial autocorrelation in socioeconomic, soil and water quality data
Lead author / co-author
I am undertaking the role of the principal data analyst of this project. The research team for this project aims to find a general trend in the relationship between spatial autocorrelation of dependent variables and the residual spatial autocorrelation from a range of research data from open data sources. Conventional machine learning algorithms are compared in multiple data settings. The results from socioeconomic data were published in geography journals such as Professional Geographer, Geographical Analysis, and Geocarto International.
Spatiotemporal shifts in the association between unemployment and mental-illness mortality in the United States before and after the Great Recession
I am working with Dr. Hui Luan to examine what spatiotemporal interaction would be the best to model the association between unemployment rates and mental-illness mortality rates in the contiguous United States. The highlights of this study is the association between unemployment and mental-illness mortality shifted through space during the study period (2001-2014), which includes the economic recession in 2001-2002 and the Great Recession (2007-2009). This study was published in a geography journal, Applied Geography.
Spatial heterogeneity of the association between PM10 and low birth weight in South Korea
I analyzed the data with more than 5.6 million neonates born between 2001-2013 in South Korea. We showed the spatially various association between PM10 pollution and low birth weight. This study was published in an environmental science journal, Environmental Research.
National model of particulate matter and nitrogen dioxide in South Korea
I processed the national-scale geodatabase to compute 324 geographic variables to develop a national model of two air pollutants (PM10 and NO2). A co-authored article published in Environmental Pollution includes more details (please see Publications)