Applying a Probabilistic Infection Model for studying contagion processes in contact networks
Introduction
A primary component of computational epidemics is modeling and simulating how infections spread in a population. Two main approaches to simulating the spread of disease are (i) stochastic agent-based modelling; and (ii) deterministic meta-population models [1], [14].
Both models assume that the individuals are in exactly one disease state. For example, the SEIR model, which we simulate in this paper, the states are Susceptible, Exposed, Infected, and Recovered. This framework is 1 is depicted in Fig. 1. S, E, I, and R represent the number of individuals in Susceptible, Exposed, Infected, and Recovered states respectively. The total population is then given by N = S + E + I + R. Parameter β is the proportion of contacts between members of S and members of E that lead to disease transmission. Parameter σ is the rate at which the exposed become infected. Parameter γ is the recovery rate at which the infected transition to the recovered state.
In stochastic agent-based models each individual (or group of individuals) in a population is represented as agents. Dyadic interactions between agents are governed by functions of the agents’ characteristics, or their environment. These interactions can be used to form a contact network. The infection spreads through the connections in this network. Meta-population models use a system of differential equations to approximate the rate of change of the number of individuals in each disease state (e.g., susceptible, infected, etc.). Here the specific connections between the individuals is not modeled.
Both these models exhibit competing benefits and drawbacks. The advantage of the stochastic agent-based approach is that it can model population heterogeneity including variations in the numbers of contacts of each individual, as well as by varying the infection parameters, such as γ, σ, per individual. The disadvantage is that due to the reliance on stochastic processes, a single run of an outbreak simulation is not representative of an expected outcome. Hundreds of repeated executions per unique set of parameters are needed in order to adequately estimate trends in the data.
In an almost exact reverse, meta-population disease models are computationally efficient due to their deterministic nature. Further, closed form approximations of significant epidemiological parameters such as the basic reproduction number R0 (i.e. the expected number of secondary cases resulting from a single infectious individual in a completely susceptible population) can be derived analytically using meta-population models. But these models do not represent the diversity of the individuals, and assume a homogeneous mixing rate within a homogeneous population. Motivated by these trade-offs, our goal is to combine the advantages of these two popular epidemiological models. We introduce the Probabilistic Infection Model (PIM), which combines the heterogeneity of the stochastic models with the computational efficiency and deterministic nature of the meta-population models. The key idea of PIM is to calculate the probabilities of the four SEIR states associated with that vertex for each vertex in a contact network.
To compute the probability function, we leverage the research conducted in escape probabilities by Thomas and Weber [32]. The probabilities for each state and each vertex are compounded over windows of time corresponding to the latent and infectious periods of the given disease. This allows for probabilistic values of different states over time at the individual levels and also provides the expected values of the sizes of the SEIR sub-populations corresponding to each state. As an added advantage, our proposed PIM allows us to compute an expression for , which yields the value of R0 for specific single infectious individuals in an otherwise susceptible contact network. In Table 1 we provide a comparison between the stochastic model, the meta-population model and our proposed PIM.
We applied our model to a contact network created from class enrollment data from the University of North Texas, as well as on two other contact networks that are available online; (i) on a network of friendship of students in high school and (ii) a network of students living in a residential hall. We conducted our experiments by simulating the following epidemics; two varieties of influenza, measles, and Covid-19. We compared simulation results as well as the timing of the PIM with those produced by the stochastic models. Our results demonstrate that the PIM simulations are similar to those produced by averaging trials from Monte Carlo models. This similarity is most notable when simulating diseases that are highly infectious, such as measles.
Section snippets
The Probabilistic Infection Model
In this section we describe our novel Probabilistic Infection Model (PIM). In Table 2, we provide a list of the terms that we use in our computations, along with their definitions. The input to both the stochastic model (SM) and the PIM is a contact network among individuals. In the SM model, a contact event is simulated by a vertex selecting a single neighbor with a given probability. Due to this inherent stochasticity of the model, the simulation must be executed multiple times to estimate
Empirical results
In this section we present our experimental results of comparing the simulation of PIM with the stochastic Monte-Carlo simulations.
Datasets used. Creating a reliable contact network is challenging in computational epidemiology [9]. This is because such as traditional methods of determining contacts such as surveys or sensor based tracking cannot scale. Surveys are also affected by recall bias, because participants may not remember all of their contacts.
As a solution to this problem, we observe
Relationship between graph structure and probability of infection
We examined the results from the Measles, Influenza 1, and Influenza 2 experiments to identify potential associations between graph structure and simulation outcomes. In each of these three experiments, a single vertex served as the initial source of infection. As before, edge weights in the graph were chosen to be proportional to pairwise contact probabilities and were derived from the total duration in hours of shared class time between a pair of vertices. For each vertex, the graph distance
Effect of varying COVID-19 transmissibility on outbreak dynamics
The PIM was used to investigate the effects of non-pharmaceutical interventions on a simulated COVID-19 outbreak within the same university population used for Measles and Influenza simulations earlier in this manuscript. Social distancing was simulated by changing how contacts were dervived from shared classtime; and mask wearing was simulated by varying the transmissibility of the disease. Due to the deterministic nature of PIM, only a single simulation run was required to generate a result
Related research
Computational epidemiology is an active area of research. Despite the advances in modeling infection spread using networks several challenges exist. As discussed in [27], including developing more accurate network models from data, extending epidemic simulations to dynamic and weighted networks, understanding how the structure of the network relates to the spread of diseases and developing prevention strategies. These challenges represent on-going problems and are being addressed by several
Conclusion and future work
In this paper, we introduce a Probabilistic Infection Model for simulating the spread of infectious diseases on contact networks. Our model encapsulates the advantages of both deterministic meta-population models as well as stochastic models on contact networks. We further propose a method of obtaining contact networks based on the scheduled activities of individuals in specific environments (e.g., businesses, schools, etc.), and simulate our model on a contact network built from a university's
Author statement
Qian developed the code and conducted the experiments. Mikler, Bhowmick, O’Neill and Ramisetty-Mikler provided the conceptual ideas and designed the experiments. All the authors participated in writing the paper.
Acknowledgement
Bhowmick's research was supported by the National Science Foundation under Grant No. 1916084.
Conflicts of interest: None declared.
William Qian is an undergraduate student in the complex systems lab at University of Pennsylvania. His research interests are in synchronization networks, dynamical systems, and temporal networks. He worked on the research published in this paper as a Texas Academy of Maths and Sciences(TAMS) student at University of North Texas.
References (34)
- et al.
Six challenges in measuring contact networks for use in modelling
Epidemics
(2015) - et al.
Exploring social structure using dynamic three-dimensional color images
Soc. Netw.
(1998 Apr) - et al.
Using the contact network model and metropolis-hastings sampling to reconstruct the covid-19 spread on the “diamond princess”
Sci. Bull.
(2020) Peer influence groups: identifying dense clusters in large networks
Soc. Netw.
(2001 Oct)- et al.
Eight challenges for network epidemic models
Epidemics
(2015) - et al.
Centre for the mathematical modelling of infectious diseases COVID-19 working group, Jit, M., Klepac, P.: the effect of control strategies to reduce social mixing on outcomes of the covid-19 epidemic in wuhan, china: a modelling study
The Lancet. Public Health.
(2020 May) - et al.
Comparing large-scale computational approaches to epidemic modeling: agent-based versus structured metapopulation models
Bmc Infect. Dis.
(2010) - et al.
Digital contact-tracing during the covid-19 pandemic: an analysis of newspaper coverage in germany, austria, and switzerland
PLOS ONE
(2021 02) - et al.
Seasonal transmission potential and activity peaks of the new influenza a (h1n1): a monte carlo likelihood analysis based on human mobility
BMC Med.
(2009) - et al.
The dynamic nature of contact networks in infectious disease epidemiology
J. Biol. Dyn.
(2010)
Epidemic models on social networks-with inference
Stat. Neerl.
Estimating influenza latency and infectious period durations using viral excretion data
Epidemics
An interactive, web-based high performance modeling environment for computational epidemiology
ACM Trans. Manage. Inf. Syst.
A method for reducing the severity of epidemics by allocating vaccines according to centrality.
ACM Conference on Bioinformatics, Computational Biology, and Health Informatics
The effect of contact investigations and public health interventions in the control and prevention of measles transmission: a simulation study
PLoS One
Using a real-world network to model localized covid-19 control strategies
Nat. Med.
Modeling targeted layered containment of an influenza pandemic in the United States
Proc. Natl. Acad. Sci.
Cited by (3)
Deep learning modeling of public's sentiments towards temporal evolution of COVID-19 transmission
2022, Applied Soft ComputingCitation Excerpt :However, the sample size is small for both analyses (i.e., 5000 Tweets and less than 100 questionnaire responses). Efforts have been paid to seek other evidence for pandemic prediction and control, such as infection probability under varying distances from the source of infection [33], visitors’ trajectory data for crowd control [34], aggregators by demographic information [35], and combining daily COVID-19 time-series records and COVID-19 related Twitter data to model and forecast the growth rate in the number of confirmed COVID-19 cases globally [36]. In summary, empirical evidence of global sentiment variations with a fine time resolution (i.e., daily) towards the evolving COVID-19 is scares.
Optimal allocation strategies for prioritized geographical vaccination for Covid-19
2022, Physica A: Statistical Mechanics and its ApplicationsCitation Excerpt :In another respect, the mathematical modeling gives epidemiologists a powerful tool in order to understand the virus propagation mechanism and consequently try to control it with different means [7–12].
Agent-Based Simulation and Modeling of COVID-19 Pandemic: A Bibliometric Analysis
2022, Journal of Disaster Research
William Qian is an undergraduate student in the complex systems lab at University of Pennsylvania. His research interests are in synchronization networks, dynamical systems, and temporal networks. He worked on the research published in this paper as a Texas Academy of Maths and Sciences(TAMS) student at University of North Texas.
Sanjukta Bhowmick is an Associate Professor in the Computer Science and Engineering department at the University of North Texas. She obtained her Ph.D. from the Pennsylvania State University. She had a joint postdoc at Columbia University and Argonne National Lab. Her current research is on understanding change in complex network analysis, with a focus on developing scalable algorithms for large dynamic networks and developing uncertainty quantification metrics for network analysis.
Susie Ramisetty-Mikler is a Research Associate Professor in Epidemiology and Biostatistics, Population Health Sciences, School of Public Health At Georgia State University, Atlanta, GA. Research & Activities of Interest: Survey design, sampling, psychometrics, and statistical analysis of large general population health data; Alcohol epidemiology, alcohol related problems including family, partner violence, depression; child/Adolescent risk behaviors, alcohol/drug use, sexual risk taking- etiology, consequences; current research - birth defects, pre-conception health.
Marty O’Neill is the Director of the Center for Computational Epidemiology and Response Analysis (CeCERA), and a Research Associate Professor in the Department of Biological Sciences at University of North Texas (UNT). He obtained his Ph.D. in Computer Science from UNT in 2014. His current projects include outbreak modeling, visualization of complex data, geospatial analysis, and response plan design.
Armin R. Mikler is a Professor and Chair of the Department of Computer Science at Georgia State University. He received his PhD in Computer Science from Iowa State University in 1995. As a professor of Computer Science and Engineering at the University of North Texas from 1997 -2020, Dr. Mikler directed the Center for Computational Epidemiology and Response Analysis (CeCERA). His research interests include Computational Epidemiology and Disaster Informatics with focus on data-driven response plan design and plan optimization.