ABSTRACT
Chapter 6 introduces the network community detection approach to delineation of hospital service areas (HSAs). A patient-to-hospital origin-destination (OD) flow matrix can be treated as a complex network, and the derivation of HSAs becomes a task of segmenting the network into subnetworks (communities). The resulting subnetworks have the maximum connections within each and the minimum connections between them, commonly referred to as “community detection” in network science. Adapting a community detection method for defining HSAs needs to account for some spatial constraints such as area adjacency and threshold population size. This chapter instigates two recently developed methods, based on two popular community detection methods, namely “spatially constrained Louvain and Leiden algorithms”. Both are customized in tools in ArcGIS Pro. A case study illustrates and compares the two methods in defining HSAs in Florida to facilitate comparison to those defined by the Dartmouth method.
