Page 34 - GIS for Science Technology Showcases

188
GIS for Science
MODELING RELATIONSHIPS
USING LINK ANALYSIS
EBOLA OUTBREAK, SIERRA LEONE 2014–2015
Linda Beale, Esri
Link analysis played a strong role in tracking the person-to-person transmission of the Ebola virus in Sierra Leone in 2014–2015. Link analysis uses graph theory for evaluating connections or relationships between nodes, where nodes can represent people, places, objects, and events. You can visualize the results of link analysis using an association matrix, or more typically, a link chart to evaluate the patterns of interest. Geographically, flow maps are used to show the movement of objects from one location to another.
Link analysis
Several different measures of topological centrality are possible with link analysis, each of which seeks to answer a slightly different question. The degree of nodes shows the measure of centrality, and normalized centrality measures adjust for network size.
Degree centrality allows you to see what is flowing through the network and identify the most influential nodes. The important nodes are identified as those having the most connections. Degree centrality can have directionality so that nodes with higher out-degree values are more central, or nodes with higher in-degree are more important. Degree centrality is a local measure that considers a node’s importance within its locality, but not any indirect relationships.
Betweenness centrality measures the extent to which a node lies on paths between other nodes. Nodes with high betweenness are likely to have an important influence within a network by virtue of their control over information passing between other nodes. Removal of nodes with high betweenness from the network will have the greatest disruption on communications or flow across that network as they lie on the largest number of paths.
Closeness centrality is based on the average of the shortest network path distance between nodes and identifies nodes as being more central if they are closer to most of the nodes in the network. Closeness centrality is used to determine which nodes are most closely associated to the other nodes in the network.
Eigenvector centrality depends on the number of neighbors and the quality of its connections, with the most central nodes being important nodes that are connected to other important nodes. Eigenvector centrality is of value to determine the nodes that are part of a cluster of influence.
Link analysis together with spatial data analysis offer enormous value for epidemiological analysis of distributions, patterns, and determinants of health and disease conditions within populations. Understanding the development of epidemics
caused by infectious diseases and the impact of interventions, together with an understanding of the geography of at-risk populations and potential transmission pathways, can help ensure effective responses in the future.
Pathways of transmission
Infectious disease epidemiology can use link analysis
to show connectivity of individuals or places. The measure of centrality allows the isolation or accessibility
to be measured. If a link directly connects two nodes, these nodes can be evaluated as transmission events from individual to individual or place to place. These relationships indicate potential transmission pathways for infections between individuals or through populations.
Interactions between microorganisms such as bacteria and
viruses cause infectious diseases. Zoonotic diseases are infectious diseases of animals that can cause disease when transmitted to humans.
A transmission network can be created using individual data of infected people linked to those from whom they caught the infection and to any others they infected. This network will show all the links through which infection spread
in the outbreak; however, it will not show interactions that led to infection transmission. Because nodes represent places at a population level, the nodes represent locations of high connectivity of infected cases, which together with population data can help define those areas where population interactions were highest.
West Africa, Ebola outbreak 2014–2015
Data from the World Health Organization shows Ebola cases in Sierra Leone from the 2014–2015 Ebola outbreak in West Africa. Ebola, a zoonotic disease, spreads in the human population through human-to-human transmission. Home, infection, and death locations for known cases show the geographic spread, with the node sizes showing the degree centrality value of those locations.
Research shows that during the 2014–2015 West Africa outbreak, the majority of transmission events occurred between family members. The link chart shows the relationships to known contacts with people diagnosed with Ebola. Understanding traditional practices and Ebola transmission pathways ultimately led to changes in behaviors related to mourning and the adoption of safe burial practices.