Continuous Models of Epidemic Spreading in Heterogeneous Dynamically Changing Random Networks

TL;DR

This paper introduces a new method using nonlinear ODEs to model epidemic spreading in complex, changing networks, offering a computationally efficient alternative to Monte-Carlo simulations, demonstrated through HIV-AIDS data.

Contribution

The paper presents a novel nonlinear ODE-based modeling approach for epidemic spread in dynamic heterogeneous networks, improving computational efficiency and analytical capabilities.

Findings

ODE model accurately predicts HIV-AIDS spread dynamics

Method outperforms Monte-Carlo simulations in efficiency

Model aligns well with historical epidemic data

Abstract

Modeling spreading processes in complex random networks plays an essential role in understanding and prediction of many real phenomena like epidemics or rumor spreading. The dynamics of such systems may be represented algorithmically by Monte-Carlo simulations on graphs or by ordinary differential equations (ODEs). Despite many results in the area of network modeling the selection of the best computational representation of the model dynamics remains a challenge. While a closed form description is often straightforward to derive, it generally cannot be solved analytically; as a consequence the network dynamics requires a numerical solution of the ODEs or a direct Monte-Carlo simulation on the networks. Moreover, Monte-Carlo simulations and ODE solutions are not equivalent since ODEs produce a deterministic solution while Monte-Carlo simulations are stochastic by nature. Despite some recent advantages in Monte-Carlo simulations, particularly in the flexibility of implementation, the computational cost of an ODE solution is much lower and supports accurate and detailed output analysis such as uncertainty or sensitivity analyses, parameter identification etc. In this paper we propose a novel approach to model spreading processes in complex random heterogeneous networks using systems of nonlinear ordinary differential equations. We successfully apply this approach to predict the dynamics of HIV-AIDS spreading in sexual networks, and compare it to historical data.