Biomedical Engineering Reference
In-Depth Information
Figure 3
.
Temporal sparseness
. Dynamic social contact is modeled by generating a fixed set of
possible contacts for each individual and realizing a constant number per unit time according to a
specified probability model. Reducing the total number of realized contacts from 100% (
A
) to 50%
(
B
) ensures survival of at least some members of the population, even in highly vulnerable popula-
tions such as this small-world network. However, the sociospatial structure of linkage realization
probabilities also has an important effect as shown in (
C
), where the total number of contacts realized
per unit time is eequivalent to (
B
), but the probability of realizing each link is inversely proportional
to the distance between individuals in social space.
3.2.1. Disease-Reactive Dynamics
The effects of intrinsic temporal variability are amplified by illness-induced
changes to social interaction. One factor that would seem to play a major role is
a healthy person's conscious avoidance of the sick, either at the behest of health
authorities or through their own spontaneous social quarantines. However, the
potential value of this mechanism is undermined by the fact that many patho-
gens are transmissible for days or even years before any signs of illness emerge
to provoke social withdrawal (e.g., upper respiratory viral infections, HIV, or the
"infectious" habit of smoking). Most visible symptoms are generated by the im-
mune response, rather than the pathogen, and thus require at least a day or two
to develop. Quarantines also demand extreme vigilance on the part of a large
number of hosts if they are to effectively protect a population, or even a specific
individual. Given the high degree of clustering in social networks, A can infect
B quite certainly by transmitting disease to their mutual friends C, D, and E, no
matter how studiously B avoids A. Thus, B's health depends on the simultaneous
diligence of C, D, and E, and all require some overt sign of disease to trigger
withdrawal from A. Figure 4 shows the results of epidemiologic simulations in
which uninfected hosts probabilistically reduce contact with infected individuals
once signs of illness appear. In this case, hosts are latently infectious for 2 time
units before becoming overtly infectious for one more time unit before dying.
Relative to a constant-contact network (4A), social systems that dynami-
cally withdraw contact from overtly sick individuals (4B) experience consider-
able advantages in population survival even when the vast majority of indi-
