Changzhen Wang, PhD Student, Wins the 2021 Health Data Visualization Award

Changzhen Wang, PhD student, won the 2021 Health Data Visualization Award (static mapping group) by the AAG Health & Medical Geography Specialty Group.  

In this project, our motivation is to delineate meaningful and reliable spatial units of Cancer Service Areas (CSAs) for appropriately assessing the effectiveness of cancer care. The CSAs can be used to study geographic variation of health care delivery, utilization, and outcomes, specific to cancer care and inform cancer care practices and policies. Supported by the National Cancer Institute (NCI) grant (PI: Dr. Tracy Onega and Dr. Fahui Wang, No.: R21CA212687), we developed the spatially constrained Leiden (ScLeiden) method to define spatially continuous CSAs. Based on our recently published paper (Wang et al., 2020), the derived CSAs have boundaries aligned well with the major service volumes of cancer patients which can be vividly visualized via a network flow map. The performance is also evaluated by several commonly used indices in health studies, such as high localization index and balanced region size. The method is reproducible, automated, scale-flexible, and efficient, demonstrating the scientific soundness and utility in capturing regional cancer care markets. The GIS tool and derived CSAs will be publicly available at, which can be used by stakeholders including health care professionals for their research purposes.

The network visualization map depicts the CSAs derived from the ScLeiden method and associated service volume flows. The map was created in Gephi and ArcGIS Pro. Gephi is an open source software for network analysis and visualization. It is very good at creating curved-line network flow maps which is not available in ArcGIS Pro. Firstly, we prepared the spatial network of cancer care utilization volumes based on 2014-15 Medicare data from Centers for Medicare and Medicaid Services (CMS) and ZIP code data from U.S. Census Bureau. The network contained a node file and an edge file. The node file referred to 5,969 ZIP code points that represented the population-weighted centroids of ZIP code areas in the nine-state Northeast region. The edge file referred to 29,875 flows from the ZIP code points of patients (origin nodes) to ZIP code points of hospitals (destination nodes). We then imported these two files into Gephi and used GeoLayout to show the geographic coordinates of ZIP points. Each node had its size and color ranked by the incoming service flows ending at the node. Each edge had the color ranked by service volumes in curved lines and its thickness was proportional to the service volumes between two nodes. We visualized and exported the spatial network as a map. Secondly, we used the GIS tool of implementing the ScLeiden method on the network to delineate 17 CSAs with a global optimal modularity. Each derived CSA has maximum internal service volumes while different CSAs have minimum service volumes. Thirdly, we added the ZIP code points layer and the polygon layer of 17 CSAs in ArcGIS Pro and symbolized CSAs using their unique values. The file of network flows was also added and georeferenced to be aligned with ZIP code points. We inserted legend, north arrow, scalar bar, and other important elements, and output the map. The detailed process is also illustrated in our upcoming new book (by Fahui Wang and Changzhen Wang).

Wang, C., Wang, F., Onega, T., 2020. Network optimization approach to delineating health care service areas: Spatially constrained Louvain and Leiden algorithms. Transactions in GIS (available online).