[Zurück]

C. Urach, G. Zauner, N. Popper, F. Miksch, P. Einzinger, G. Endel, I. Schiller-Frühwirth, F. Breitenecker:
"Evaluation of dynamic modelling approaches for vaccination strategies";
Poster: 4th Vaccine and ISV Annual Global Congress, Vienna, Austria; 03.10.2010 - 05.10.2010.

PURPOSE: Several dynamical approaches can simulate epidemics and vaccination strategies. Generally the models can be divided into top-down approaches like Markov models and differential equations and bottom-up approaches like cellular automata and agent based models. Top-down approaches are characterized by cumulative values that are representing groups of people. Bottom-up approaches in contrast consider individuals. Both approaches have advantages and disadvantages. Top-down approaches can be analyzed very well with mathematical methods while bottom-up approaches require comparison of the outcome of simulation runs with different parameter sets. To improve validity of model structures a method that compares different approaches for epidemic models is introduced. METHODS: Statistical calculations and Markov models are static while other approaches like differential equations or individual based models are dynamic. In this context dynamic does not only stand for a simulation over a time but also for a model where the calculation of the next time step or period depend on the current state of the model. Since the transition matrices in Markov models are calculated before execution time it is considered not to be dynamic. The advantage of dynamic models is that they can produce highly nonlinear behaviour that cannot be reached with static calculations. To validate the structure of such nonlinear models different model types are implemented and compared. Results are compared; sensitivity analysis is done separately. RESULTS: Outcome of vaccination against streptococcus pneumoniae was tested. A differential equations model and an agent based model could reproduce results of published Markov models. As soon as we consider population dynamics, herd immunity and serotype replacement the Markov model was not able to fulfil the structural requirements any more while dynamic approaches still work. CONCLUSION: Dynamic models offer more information and opportunities for epidemic simulation. Usage of different approaches provides at least comparable reliability.