Research On Modeling Of Infectious Disease Spreading Based On Complex Network Theory

The spread of epidemics in human populations is a typical public health incident, and has also become one of the major public safety issues that humanity is facing in the 21st century. Traditional epidemiological research has focused on the use of rate-based differential-equations to establish the mathematical models of the spread of infectious diseases. Although successful in explaining certain phenomena of epidemic spreading, the traditional approach is unable to deal with epidemic spreading in realistic large-scale social contact networks with significant heterogeneity due to its assumption of uniformly mixing populations. Therefore, the approach based on a combination of complex network theory and epidemiology has become an important trend in modeling of transmission of infectious disease. With these facts, in this paper, exploratory studies that focused on the evolution behavior, formation mechanism and structural characteristics of social contact networks, as well as the impact of individual behavior patterns on the spread of infectious disease have been carried out from both theoretical and applied levels.The main contributions of theoretical research are as follows: (1) introduction of an alternate social network model - the crowd in time alternatively form the family network and public network that are constituted by household structures and community structures respectively, to simulate social contact network's evolution process which is driven by daily activities of individuals. The results show that the propagation velocity of infectious diseases in the family network is significantly lower than that in the public network. (2) Proposing a growing social network model incorporating local structures (basic social structures such as household, roommates, etc.). It has been found that the introduction of local structures makes the model have some major features of social networks, including a power law degree distribution with an adjustable exponent, nontrivial clustering, and assortative mixing between degrees of adjacent vertices. The infected population grows in an exponential form and the propagation speed increases markedly with the increasing size of local structure. In addition, the existence of local structure produces the convergence properties of high-order moment of the degree distribution function, leading to the network with nontrivial epidemic threshold. (3) Establishing a spatial metapopulation network model to describe the spatial structure of populations. The quantitative knowledge of human traveling statistics is first introduced into the epidemic model. The results show that restricting the characteristic travel distance, shortening the characteristic waiting time or increasing the returning probability can effectively reduce the occurrrence probability of global outbreaks.?In applied studies, we have developed an individual-based simulation model of large-scale spread of infectious diseases, combined with the actual transportation networks of railway passenger transport and civil aviation. The model is split into a population structure model, a human travelling model and a local stochastic SEIR (Susceptible-Exposed-Infectious-Removed) epidemic model. Then an application of the model was illustrated for a case study of China's SARS epidemic in 2003, in which the simulation results are basically consistent with the historical data, proving the feasibility of the model.