Inference for Network Parameters in Stochastic Epidemic Models

Please enable javascript in your browser.

Dehbod, Siamak | 2014

677 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 46532 (04)
University: Sharif University of Technology
Department: Physics
Advisor(s): Ejtehadi, Mohammad Reza
Abstract:
Different models describing the procedure of epidemics are complicated and involving numerous factors. Some of these models that used a stochastic network for modelling epidemics had been successful in statistical prediction of spreading epidemics. However, in many cases there is no direct way to determine parameters of these models. In this thesis it has been desired to estimate those parameters based on historical data of epidemics To estimate structural parameters, concerning epidemic network, two independent inference have been studied. In the first inference, the probability of disease transmission from one person to another is estimated.
In the second, using Bayesian inference, distribution of the parameters of epidemic models is estimated in a society with stochastic structure. The first inference, while having numerous samples, can estimate epidemic parameters very well. As increasing samples, the accuracy of estimating parameters grow. Considering the result of the second inference proved that this inference only has utility for analyzing correlation of model parameters and comparing different groups of data
Keywords:
Random Networks ; Inverse Solution ; Epidemic ; Social Networks ; Parameter Estimation