Statistical Inference Framework for Source Detection of Contagion Processes on Arbitrary Network Structures

Abstract

In this paper we introduce a statistical inference framework for estimating the contagion source from a partially observed contagion spreading process on an arbitrary network structure. The framework is based on a maximum likelihood estimation of a partial epidemic realization and involves large scale simulation of contagion spreading processes from the set of potential source locations. We present a number of different likelihood estimators that are used to determine the conditional probabilities associated to observing partial epidemic realization with particular source location candidates. This statistical inference framework is also applicable for arbitrary compartment contagion spreading processes on networks. We compare estimation accuracy of these approaches in a number of computational experiments performed with the SIR (susceptible-infected-recovered), SI (susceptible-infected) and ISS (ignorant-spreading-stifler) contagion spreading models on synthetic and real-world complex networks.The structure of vast majority of biological networks (biochemical, ecological), technological networks (internet, transportation, power grids), social networks and information networks (citation, WWW) can be represented by complex networks [16], [7], [3]. Epidemic or contagion processes are amongst the most prevalent type of dynamic processes of interest characteristic for these real-life complex networks and they include disease epidemics, computer virus spreading, information and rumor propagation [23]. Different mathematical frameworks have been used to study epidemic spreading. We can divide them into two major categories based upon assumptions they make: the homogeneous mixing framework and the heterogeneous mixing framework. The homogeneous mixing framework assumes that all individuals in a population have an equal probability of contact. This is a traditional mathematical framework [12], [10] in which differential equations are used to model epidemic dynamics. The heterogeneous mixing framework assumes * Corresponding author. Adress: Rudjer Bošković Institute,

Identification of Patient Zero in Static and Temporal Networks: Robustness and Limitations

“…The second motivation comes from silent spreading of a certain class of computer viruses and worms through computer networks which become active simultaneously on a specific date. Unlike other approaches [12][13][14][15][16][17][18][19][20][21], we identified different source detectability regimes and our methodology is applicable to arbitrary network structures, and is limited solely by the ability to computationally produce realizations of the particular contagion process.…”

“…Due to its practical aspects and theoretical importance, the epidemic source detection problem on contact networks has recently gained a lot of attention in the complex network science community. This has led to the development of many different source detection estimators for static networks, which vary in their assumptions on the network structure (locally tree-like) or on the spreading process compartmental models (SI, SIR) [12][13][14][15][16][17][18][19][20][21] or both.…”

Bayesian Inference of Epidemics on Networks via Belief Propagation

“…Several studies have recently proposed maximum likelihood estimators based on various kinds of information: topological centrality [2][3][4], measures of the distance between observed data and the typical outcome of propagations from given initial conditions [5], or the estimation of the single most probable path [6]. Other estimators are derived under strong simplifying assumptions on the graph structure or on the spreading process [7,8].…”

Inference of causality in epidemics on temporal contact networks

“…In recent years, this problem has received considerable attention, especially on discrete time models12345. For these models, we recently proposed an approximate Bayesian method based on Belief Propagation (BP)67, that gave the first exact tractable solution to a family of discrete time inference problems on acyclic graphs and an excellent approximation on general graphs, including real ones.…”

“…Identifying past features of an epidemic outbreak remains a challenging problem even for simple stochastic epidemic models, such as the susceptible-infected (SI) model and the susceptible-infected-recovered (SIR) model. In recent years, this problem has received considerable attention, especially on discrete time models 1 2 3 4 5 . For these models, we recently proposed an approximate Bayesian method based on Belief Propagation (BP) 6 7 , that gave the first exact tractable solution to a family of discrete time inference problems on acyclic graphs and an excellent approximation on general graphs, including real ones.…”