Sets the epidemic parameters for stochastic network models
simulated with netsim.
param.net(
inf.prob,
inter.eff,
inter.start,
act.rate,
rec.rate,
a.rate,
ds.rate,
di.rate,
dr.rate,
inf.prob.g2,
rec.rate.g2,
a.rate.g2,
ds.rate.g2,
di.rate.g2,
dr.rate.g2,
...
)Probability of infection per transmissible act between a susceptible and an infected person. In two-group models, this is the probability of infection to the group 1 nodes. This may also be a vector of probabilities, with each element corresponding to the probability in that time step of infection (see Time-Varying Parameters below).
Efficacy of an intervention which affects the per-act probability of infection. Efficacy is defined as 1 - the relative hazard of infection given exposure to the intervention, compared to no exposure.
Time step at which the intervention starts, between 1 and
the number of time steps specified in the model. This will default to
1 if the inter.eff is defined but this parameter is not.
Average number of transmissible acts per partnership
per unit time (see act.rate Parameter below). This may also be
a vector of rates, with each element corresponding to the rate in in
that time step of infection (see Time-Varying Parameters below).
Average rate of recovery with immunity (in SIR models)
or re-susceptibility (in SIS models). The recovery rate is the
reciprocal of the disease duration. For two-group models, this is the
recovery rate for mode 1 persons only. This parameter is only used for
SIR and SIS models. This may also be a vector
of rates, with each element corresponding to the rate in that time
step of infection (see Time-Varying Parameters below).
Arrival or entry rate. For one-mode models, the arrival rate is
the rate of new arrivals per person per unit time. For two-group
models, the arrival rate may be parameterized as a rate per mode 1
person time (with group 1 persons representing females), and with the
a.rate.g2 rate set as described below.
Departure or exit rate for susceptible. For two-group models, it is the rate for the group 1 susceptible only.
Departure or exit rate for infected. For two-group models, it is the rate for the group 1 infected only.
Departure or exit rate for recovered. For two-group models, it
is the rate for the group 1 recovered only. This parameter is only
used for SIR models.
Probability of transmission given a transmissible act between a susceptible group 2 person and an infected group 1 person. It is the probability of transmission to group 2 members.
Average rate of recovery with immunity (in SIR
models) or re-susceptibility (in SIS models) for group 2
persons. This parameter is only used for two-group SIR and
SIS models.
Arrival or entry rate for group 2. This may either be
specified numerically as the rate of new arrivals per group 2 persons
per unit time, or as NA in which case the mode 1 rate,
a.rate, governs the group 2 rate. The latter is used when, for
example, the first group is conceptualized as female, and the female
population size determines the arrival rate. Such arrivalss are evenly
allocated between the two groups.
Departure or exit rate for group 2 susceptible.
Departure or exit rate for group 2 infected.
Departure or exit rate for group 2 recovered. This
parameter is only used for SIR model types.
Additional arguments passed to model.
A key difference between these network models and DCM/ICM classes is the
treatment of transmission events. With DCM and ICM, contacts or partnerships
are mathematically instantaneous events: they have no duration in time, and
thus no changes may occur within them over time. In contrast, network models
allow for partnership durations defined by the dynamic network model,
summarized in the model dissolution coefficients calculated in
dissolution_coefs. Therefore, the act.rate parameter has
a different interpretation here, where it is the number of transmissible acts
per partnership per unit time.
The inf.prob, act.rate, rec.rate arguments (and their
.g2 companions) may be specified as time-varying parameters by passing
in a vector of probabilities or rates, respectively. The value in each
position on the vector then corresponds to the probability or rate at that
discrete time step for the infected partner. For example, an inf.prob
of c(0.5, 0.5, 0.1) would simulate a 0.5 transmission probability for
the first two time steps of a person's infection, followed by a 0.1 for the
third time step. If the infected person has not recovered or exited the
population by the fourth time step, the third element in the vector will
carry forward until one of those events occurs or the simulation ends. For
further examples, see the NME Course
Tutorials.
In addition to deterministic parameters in either fixed or time-varying
varieties above, one may also include a generator for random parameters.
These might include a vector of potential parameter values or a statistical
distribution definition; in either case, one draw from the generator would
be completed per individual simulation. This is possible by passing a list
named random.params into param.net, with each element of
random.params a named generator function. See the help page and
examples in generate_random_params. A simple factory function
for sampling is provided with param_random but any function
will do.
To build original models outside of the base models, new process modules
may be constructed to replace the existing modules or to supplement the
existing set. These are passed into the control settings in
control.net. New modules may use either the existing model
parameters named here, an original set of parameters, or a combination of
both. The ... allows the user to pass an arbitrary set of original
model parameters into param.net. Whereas there are strict checks with
default modules for parameter validity, these checks are the user's
responsibility with new modules.
param.net sets the epidemic parameters for the stochastic network
models simulated with the netsim function. Models
may use the base types, for which these parameters are used, or new process
modules which may use these parameters (but not necessarily). A detailed
description of network model parameterization for base models is found in
the Basic Network Models tutorial.
For base models, the model specification will be chosen as a result of
the model parameters entered here and the control settings in
control.net. One-group and two-group models are available,
where the latter assumes a heterogeneous mixing between two distinct
partitions in the population (e.g., men and women). Specifying any two-group
parameters (those with a .g2) implies the simulation of a two-group
model. All the parameters for a desired model type must be specified, even if
they are zero.
Use init.net to specify the initial conditions and
control.net to specify the control settings. Run the
parameterized model with netsim.
# NOT RUN {
## Example SIR model parameterization with fixed and random parameters
# Network model estimation
nw <- network_initialize(n = 100)
formation <- ~edges
target.stats <- 50
coef.diss <- dissolution_coefs(dissolution = ~offset(edges), duration = 20)
est <- netest(nw, formation, target.stats, coef.diss, verbose = FALSE)
# Random epidemic parameter list
my_randoms <- list(
act.rate = param_random(1:3),
inf.prob = function() rbeta(1, 1, 2)
)
# Parameters, initial conditions, and control settings
param <- param.net(rec.rate = 0.02, random.params = my_randoms)
init <- init.net(i.num = 10, r.num = 0)
control <- control.net(type = "SIR", nsteps = 100, nsims = 5,
resimulate.network = FALSE, tergmLite = FALSE)
# Simulate the model
sim <- netsim(est, param, init, control)
# Print and plot
sim
plot(sim)
# }
Run the code above in your browser using DataLab