Analyzing the Patterns of Space-Time Distances for Tracking the Diffusion of an Epidemic - Space-Time Integration in Geography and GIScience - page 269

Fig. 15.3 Space-time
distances of the hypothetical
epidemic
The diagram of space-time distances, derived by counting RSD and RTD
(Eq. 15.6 ) from the two-dimensional matrix, can be constructed as shown in
Fig. 15.3 . A higher value in the diagram indicates a greater number of cumulative
pairs of space-time points. In this example, we used outlier detection to find the
significant large number of cumulative space-time pairs of points to identify space-
time clustering. We then selected these space-time pairs for further analysis.
Figure 15.4 shows the structure of a network constructed from selected space-
time pairs. There are two components; therefore, the spread of this epidemic
could be characterized as relocation-dependent. In the larger component, nodes
with higher degree centrality values (nodes 3, 4, 5 and 6) could be the infection
sources; the infection then diffused to nodes 8, 9 and 10. Node 4 has the highest
network-clustering coefficient, indicating that this area has the highest transmission
risk (Table 15.2 ). Thus, the spread of infection could be identified within the
component.
Through the methods we proposed, the diffusion of a hypothetical epidemic
can be depicted clearly. The areas with higher degree centrality represent probable
sources of infection, and the nodes with higher network clustering represent areas
with the greatest transmission risk. Later in time in the model, the epidemic spread to
neighboring areas (contagious diffusion). Late in the epidemic, the disease spread to
areas farther away, leaving the area that had been the source of infection (relocation
diffusion).