Two
pandemic simulation approaches are known: the multi-agent simulation
model and the differential equation model. The multi-agent model
can deal with detailed simulations under a variety of initial and
boundary conditions with standard social network models; however,
the computing cost is high. The differential equation model can
quickly deal with simulations for homogeneous populations with
simultaneous ordinary differential equations and a few parameters;
however, it lacks versatility in its use. We propose a new method
named the MADE which is a combination of these two models, such
that we use the multi-agent model in the early stage in a simulation
to determine the parameters that can be used in the differential
equation model, and then use the differential equation model in
the subsequent stage. With this method, we may deal with pandemic
simulations for real social structures with lower computing costs.
Contrary to the statistical inference method which could not predict
the final stage unless abundant information is included, the MADE
have a possibility to do that only with the earlier stage information.
The newly emerged pandemic, the novel influenza A(H1N1) case in 2009,
is dealt with. |
|
|
|
|
| MADE,
SEIR, MAS, pandemic, novel influenza A(H1N1), truncated model,
Runge-Kutta. |
|
|