Geography Reference
In-Depth Information
The first space-time pairs clustered within 100-300 m and 1-4 days; the second
clustered within 1.3-1.8 km and 1-20 days; the third clustered within 2.6-3.9 km
and 1-20 days; and the last clustered within 4.6-5.5 km and 1-20 days. In other
words, in the first two space-time clusters, the average interval between the onset of
illness for any two cases was less than 20 days and the residential distance between
the two cases was less than 2 km. Thus, most cases could have been infected due
to neighborhood characteristics. On the other hand, the residential distance between
the last two space-time clusters was in the range 2.6-5.5 km. The spread of infection
across this distance could be attributed to population mobility.
We then mapped four major clusters of space-time pairs. The network density
of these pairs is shown in Fig.
15.8
. The intense spots in Cluster A and Cluster B,
which has smaller space-time distances, are located in south Kaohsiung and at the
border between the two cities, respectively. Cluster C and Cluster D, which has a
larger range of space-time distances, were scattered across the study area, but intense
spots were also centered on the border between the two cities. The results suggest
that the infected cases in the border have intense spatial-temporal interactions.
Two network indicators to measure diffusion risk are summarized in Table
15.3
.
The average degree centrality reflects the connections of one infected person. The
results showed that Cluster C has the highest degree centrality. Table
15.4
also
shows that most of the cases in Cluster C occurred in Period 3 of the epidemic.
These results suggest that Period 3 is the transition stage from local to large-scale
transmission and that infected cases with higher centrality in Cluster C could be
the sources of large-scale transmission. The other indicator, the network-clustering
coefficient, reflects the risk of transmission. Cluster A has the highest values of
this indicator and occurred in the first period of the epidemic (Table
15.5
). The
results suggest that areas of Cluster A could be where infected persons initiated
local transmission.
In summary, the diffusion patterns show that the epidemic originated in south
Kaohsiung (Cluster A). Infections in the initial stage resulted from neighborhood
characteristics, such as vacant lots or houses with poor garbage disposal. The
epidemic followed the “contiguous diffusion” pattern and gradually spread to the
border between the two cities in Period 2 (Cluster B). Possibly due to demographic
factors, such as population mobility, the scale of epidemic followed the “relocation
diffusion” pattern and spread across a larger range of space-time distances in Periods
3and4(ClusterCandD).
15.5
Conclusions
Space and time are two important dimensions in describing epidemic dynamics
and risk characteristics. This study applied space-time distances to determine net-
work indicators that could differentiate among contagious and relocation diffusion
patterns to model epidemic progression. Using the dates of onset and residential
locations of infected persons, the proposed method also identified possible sources
Search WWH ::
Custom Search