Learn R Programming

A flexible and efficient framework for data-driven stochastic disease spread simulations

SimInf

The package provides an efficient and very flexible framework to conduct data-driven epidemiological modeling in realistic large scale disease spread simulations. The framework integrates infection dynamics in subpopulations as continuous-time Markov chains using the Gillespie stochastic simulation algorithm and incorporates available data such as births, deaths and movements as scheduled events at predefined time-points. Using C code for the numerical solvers and 'OpenMP' (if available) to divide work over multiple processors ensures high performance when simulating a sample outcome. One of our design goals was to make the package extendable and enable usage of the numerical solvers from other R extension packages in order to facilitate complex epidemiological research. The package contains template models and can be extended with user-defined models.

Getting started

You can use one of the predefined compartment models in SimInf, for example, SEIR. But you can also define a custom model 'on the fly' using the model parser method mparse. The method takes a character vector of transitions in the form of X -> propensity -> Y and automatically generates the C and R code for the model. The left hand side of the first arrow (->) is the initial state, the right hand side of the last arrow (->) is the final state, and the propensity is written between the two arrows. The flexibility of the mparse approach allows for quick prototyping of new models or features. To illustrate the mparse functionality, let us consider the SIR model in a closed population i.e., no births or deaths. Let beta denote the transmission rate of spread between a susceptible individual and an infectious individual and gamma the recovery rate from infection (gamma = 1 / average duration of infection). It is also possible to define variables which can then be used in calculations of propensities or in calculations of other variables. A variable is defined by the operator <-. Using a variable for the size of the population, the SIR model can be described as:

library(SimInf)

transitions <- c("S -> beta*S*I/N -> I",
                 "I -> gamma*I -> R",
                 "N <- S+I+R")
compartments <- c("S", "I", "R")

The transitions and compartments variables together with the constants beta and gamma can now be used to generate a model with mparse. The model also needs to be initialised with the initial condition u0 and tspan, a vector of time points where the state of the system is to be returned. Let us create a model that consists of 1000 replicates of a population, denoted a node in SimInf, that each starts with 99 susceptibles, 5 infected and 0 recovered individuals.

n <- 1000
u0 <- data.frame(S = rep(99, n), I = rep(5, n), R = rep(0, n))

model <- mparse(transitions = transitions,
                compartments = compartments,
                gdata = c(beta = 0.16, gamma = 0.077),
                u0 = u0,
                tspan = 1:150)

To generate data from the model and then print some basic information about the outcome, run the following commands:

result <- run(model)
result
#> Model: SimInf_model
#> Number of nodes: 1000
#> Number of transitions: 2
#> Number of scheduled events: 0
#> 
#> Global data
#> -----------
#>  Parameter Value
#>  beta      0.160
#>  gamma     0.077
#> 
#> Compartments
#> ------------
#>      Min. 1st Qu. Median   Mean 3rd Qu.   Max.
#>  S   1.00   19.00  30.00  40.74   60.00  99.00
#>  I   0.00    0.00   4.00   6.87   11.00  47.00
#>  R   0.00   28.00  67.00  56.39   83.00 103.00

There are several functions in SimInf to facilitate analysis and post-processing of simulated data, for example, trajectory, prevalence and plot. The default plot will display the median count in each compartment across nodes as a colored line together with the inter-quartile range using the same color, but with transparency.

plot(result)

Most modeling and simulation studies require custom data analysis once the simulation data has been generated. To support this, SimInf provides the trajectory method to obtain a data.frame with the number of individuals in each compartment at the time points specified in tspan. Below is the first 10 lines of the data.frame with simulated data.

trajectory(result)
#>    node time  S I R
#> 1     1    1 98 6 0
#> 2     2    1 98 6 0
#> 3     3    1 98 6 0
#> 4     4    1 99 5 0
#> 5     5    1 97 7 0
#> 6     6    1 98 5 1
#> 7     7    1 99 5 0
#> 8     8    1 99 5 0
#> 9     9    1 97 7 0
#> 10   10    1 97 6 1
...

Finally, let us use the prevalence method to explore the proportion of infected individuals across all nodes. It takes a model object and a formula specification, where the left hand side of the formula specifies the compartments representing cases i.e., have an attribute or a disease and the right hand side of the formula specifies the compartments at risk. Below is the first 10 lines of the data.frame.

prevalence(result, I ~ S + I + R)
#>    time prevalence
#> 1     1 0.05196154
#> 2     2 0.05605769
#> 3     3 0.06059615
#> 4     4 0.06516346
#> 5     5 0.06977885
#> 6     6 0.07390385
#> 7     7 0.07856731
#> 8     8 0.08311538
#> 9     9 0.08794231
#> 10   10 0.09321154
...

Learn more

See the vignette to learn more about special features that the SimInf R package provides, for example, how to:

  • use continuous state variables

  • use the SimInf framework from another R package

  • incorporate available data such as births, deaths and movements as scheduled events at predefined time-points.

Installation

You can install the released version of SimInf from CRAN

install.packages("SimInf")

or use the remotes package to install the development version from GitHub

library(remotes)
install_github("stewid/SimInf")

We refer to section 3.1 in the vignette for detailed installation instructions.

Authors

In alphabetical order: Pavol Bauer , Robin Eriksson , Stefan Engblom , and Stefan Widgren (Maintainer)

Any suggestions, bug reports, forks and pull requests are appreciated. Get in touch.

Citation

If you use SimInf in your research, please cite:

Widgren S, Bauer P, Eriksson R, Engblom S (2019). SimInf: An R Package for Data-Driven Stochastic Disease Spread Simulations. Journal of Statistical Software, 91(12), 1–42. https://doi.org/10.18637/jss.v091.i12

Bauer P, Engblom S, Widgren S (2016). Fast event-based epidemiological simulations on national scales. International Journal of High Performance Computing Applications, 30(4), 438–453. https://doi.org/10.1177/1094342016635723

Acknowledgments

This software has been made possible by support from the Swedish Research Council within the UPMARC Linnaeus center of Excellence (Pavol Bauer, Robin Eriksson, and Stefan Engblom), the Swedish Research Council Formas (Stefan Engblom and Stefan Widgren), the Swedish Board of Agriculture (Stefan Widgren), the Swedish strategic research program eSSENCE (Stefan Widgren), and in the framework of the Full Force project, supported by funding from the European Union’s Horizon 2020 Research and Innovation programme under grant agreement No 773830: One Health European Joint Programme (Stefan Widgren).

Versioning

The SimInf package uses semantic versioning.

License

The SimInf package is licensed under the GPLv3.

Copy Link

Version

Install

install.packages('SimInf')

Monthly Downloads

272

Version

11.1.0

License

GPL-3

Issues

0

Pull Requests

0

Stars

39

Forks

16

Maintainer

Stefan Widgren

Last Published

September 8th, 2026

Functions in SimInf (11.1.0)

as.data.frame.SimInf_individual_events

Coerce a SimInf_individual_events object to a data.frame
as.data.frame.SimInf_pmcmc

Coerce a SimInf_pmcmc object to a data.frame
abc

Approximate Bayesian computation
add_spatial_coupling_to_ldata

Add spatial coupling information to local data
SimInf_model

Create a SimInf_model object
events_SIS

Example event data for the SIS model with cattle herds
SimInf_model-class

Class SimInf_model
events_SISe3

Example event data for the SISe3 model with cattle herds
distance_matrix

Create a distance matrix between nodes for spatial models
continue_pmcmc

Continue PMCMC from an Existing Chain
indegree

Determine in-degree for each node in a model
get_individuals

Extract individuals from SimInf_individual_events
boxplot,SimInf_model-method

Box plot of number of individuals in each compartment
continue_abc

Run more generations of ABC SMC
edge_properties_to_matrix

Convert an edge list with properties to a matrix
events

Extract the scheduled events from a SimInf_model object
as.data.frame.SimInf_events

Coerce a SimInf_events object to a data.frame
as.data.frame.SimInf_abc

Coerce a SimInf_abc object to a data.frame
logLik,SimInf_pfilter-method

Extract Log-Likelihood
mparse

Model parser to define new models for SimInf
outdegree

Determine out-degree for each node in a model
individual_events

Individual events
pairs,SimInf_model-method

Scatterplot matrix of number of individuals in each compartment
lambertW0

Lambert W0 function
nodes

Example data with spatial distribution of nodes
node_events

Transform individual events to node events for a model
gdata<-

Set a global data parameter for a SimInf_model object
ldata

Extract local data from a node
SISe_sp-class

Class SISe_sp
SimInf_pfilter-class

Class SimInf_pfilter
SimInf_pmcmc-class

Class SimInf_pmcmc
events_SEIR

Example event data for the SEIR model with cattle herds
events_SIR

Example event data for the SIR model with cattle herds
pmcmc

Particle Markov chain Monte Carlo (PMCMC) algorithm
prevalence,SimInf_pfilter-method

Extract prevalence from running a particle filter
prevalence,SimInf_pmcmc-method

Extract prevalence from fitting a PMCMC algorithm
pfilter

Bootstrap particle filter
gdata

Extract global data from a SimInf_model object
package_skeleton

Create a package skeleton from a SimInf_model
prevalence,SimInf_model-method

Calculate prevalence from a model object with trajectory data
plot,SimInf_pfilter-method

Diagnostic plot of a particle filter object
prevalence

Generic function to calculate prevalence from trajectory data
punchcard<-

Set a sparse recording template for simulation results
plot,SimInf_individual_events-method

Display the distribution of individual events over time
length,SimInf_pmcmc-method

Length of the MCMC chain
n_nodes

Determine the number of nodes in a model
summary,SimInf_events-method

Detailed summary of a SimInf_events object
show,SimInf_pmcmc-method

Brief summary of a SimInf_pmcmc object
summary,SimInf_abc-method

Detailed summary of a SimInf_abc object
show,SimInf_pfilter-method

Brief summary of a SimInf_pfilter object
n_compartments

Determine the number of compartments in a model
plot,SimInf_abc-method

Display the ABC posterior distribution
n_generations

Determine the number of generations in an ABC analysis
plot,SimInf_pmcmc-method

Display the PMCMC posterior distribution
shift_matrix<-

Set the shift matrix for a SimInf_model object
shift_matrix

Extract the shift matrix from a SimInf_model object
n_replicates

Determine the number of replicates in a model
plot,SimInf_model-method

Display the outcome from a simulated trajectory
plot,SimInf_events-method

Display the distribution of scheduled events over time
show,SimInf_abc-method

Brief summary of a SimInf_abc object
set_num_threads

Specify the number of threads that SimInf should use
select_matrix

Extract the select matrix from a SimInf_model object
show,SimInf_individual_events-method

Brief summary of a SimInf_individual_events object
summary,SimInf_individual_events-method

Detailed summary of a SimInf_individual_events object
show,SimInf_events-method

Brief summary of a SimInf_events object
summary,SimInf_model-method

Detailed summary of a SimInf_model object
trajectory,SimInf_pfilter-method

Extract filtered trajectory from running a particle filter
u0<-

Update the initial compartment state (u0) in each node
trajectory

Generic function to extract data from a simulated trajectory
show,SimInf_model-method

Brief summary of a SimInf_model object
run

Run a SimInf model simulation
trajectory,SimInf_model-method

Extract data from a simulated trajectory
select_matrix<-

Set the select matrix for a SimInf_model object
u0_SIR

Example initial population data for the SIR model
u0

Get the initial compartment state (u0) in each node
u0_SEIR

Example initial population data for the SEIR model
summary,SimInf_pmcmc-method

Detailed summary of a SimInf_pmcmc object
u0_from_individual_events

Derive the initial compartment state from individual events
summary,SimInf_pfilter-method

Detailed summary of a SimInf_pfilter object
u0_SISe3

Example initial population data for the SISe3 model
u0_SIS

Example initial population data for the SIS model
v0<-

Update the initial continuous state (v0) in each node
SIS-class

Class SIS
SISe

Create an SISe model
SIR-class

Class SIR
SISe-class

Class SISe
SEIR-class

Class SEIR
SEIR

Create an SEIR model
SIR

Create an SIR model
SIS

Create an SIS model
C_code

Extract the C code from a SimInf_model object
SISe3-class

Class SISe3
SISe_sp

Create an SISe_sp model
SISe3

Create a SISe3 model
SimInf_individual_events-class

Class SimInf_individual_events
SimInf_events

Create a SimInf_events object
SimInf_abc-class

Class SimInf_abc
SimInf

A Framework for Data-Driven Stochastic Disease Spread Simulations
SimInf_events-class

Class SimInf_events
SISe3_sp

Create an SISe3_sp model
SISe3_sp-class

Class SISe3_sp