Fig 1.
Preliminary model with symmetric movement (left), biased random walk foraging model with social clustering effects (middle) and the preventive behavior model with dynamics of public awareness (right).
Table 1.
Event types, state transitions, and corresponding transition rates of the preliminary agent-based symmetric random walk model.
Table 2.
Event types, parameter values, sources and references, and corresponding physical meanings in the mathematical model for the preliminary agent-based symmetric random walk model.
Fig 2.
Simulations of symmetric random walk model.
Panels (a.1)-(d.1) compare . Increasing the rate
of agent movement from 0.25 to 0.5 to 1 to 2 leads to a heightened disease outbreak, and thus escalate severeness of the first episode of the epidemic.
Fig 3.
The continuum ODE is better approximated by the corresponding agent-based unbiased random walk model when the mobility speed further increases.
Parameters and initial data are the same as those used in Fig 2(a.1), 2(b.1), 2(c.1) and 2(d.1), except for that is changed to 10 (left) or 100 (right). Note that ℓ is still fixed as 1/100. The solid black line represents the average of eighty randomly generated paths of the agent-based model, and the red dashed line represents the outcome of the continuum ODEs.
Fig 4.
Simulations of biased random walk model.
Panels (a.1)-(d.1) compare . Increasing the spread rate
of popularity variable from 2 to 4 to 8 to 16 suppresses the spatial heterogeneity, so that aggregation clusters of the susceptible decrease in number and size. As a consequence, the spread of the epidemic accelerates.
Table 3.
Event types, parameter values and corresponding physical meanings for the biased random walk model.
Table 4.
Event types, parameter values and corresponding physical meanings for the preliminary agent-based symmetric random walk model.
Fig 5.
Simulations of Scenario I: Integrating awareness into biasness of random walks.
Panels (a.1)-(d.1) compare , where the reference is the biased random walk model without awareness in Section 2.4. Incorporated awareness in biasness alone does not necessarily constrain outbreaks. When the increment of public awareness
increases from 1 to 3 to 10 to 20, there is a high chance that the susceptible agents have already transitioned into the exposed or asymptomatic infectious ones. As a result, disease transmissions are boosted and outbreaks are escalated.
Fig 6.
Simulations of Scenario II-i: Locally-supervised disease awareness.
Panels (a.1)-(d.1) compare , where the reference is the biased random walk model without awareness in Section 2.4. Disease control is achieved if the whole community takes response in a collaborative way. When the increment of public awareness
increases from 1 to 3 to 10 to 20, the agent mobility is restrained. As a result, the disease peaks are flattened and an evident suppressed peak of the first wave is observed.
Fig 7.
Simulations of Scenario II-ii: Spatially-uniform awareness.
Panels (a.1)-(d.1) compare , where the reference is the biased random walk model without awareness in Section 2.4. When the increment of public awareness
increases from 1 to 3 to 10 to 20, a coordinated disease control campaign is implemented in the whole region. As a result, the first epidemic peak is are strongly delayed and reduced.