transreg: Parametric regression models for infectious disease...
In ekenah/transtat: Statistical Methods for Infectious Disease Transmission

Description Usage Arguments Details Value Author(s) References

Fits accelerated failure time models for infectious disease transmission using right-censored and/or left-truncated data on contact intervals in ordered pairs of individuals and infectious contact from external sources with individuals.

transreg(
  formula,
  sus,
  data,
  subset = NULL,
  na.action,
  dist = "weibull",
  xdist = dist,
  init = NULL,
  fixed = NULL,
  optim_method = "BFGS",
  ...
)

`formula`	A formula of the form "response ~ terms". The response must be an object returned by `Surv`. The formula terms can include an `ext` term giving an external row indicator.
`sus`	A name (without quotes) or character string (with quotes) giving the name of the variable in `data` that contains the susceptible member of each pair.
`data`	A data frame containing the variables named in `formula`.
`subset`	An expression indicating which rows of `data` should be included in the model fit. In stats::model.frame, subsetting is done after evaluating the formula, which can lead to empty warnings if time == 0 or start >= stop in data rows not included in the subset. This can be avoided by subsetting the data in the data argument itself.
`na.action`	A missing-data filter applied to `data` after `subset`. Defaults to `options()$na.action`.
`dist`	A string partially matching a survival time distribution from `transreg.distributions` to specify the internal contact interval distribution. Current options are `exponential` (the default), `weibull`, and `loglogistic`.
`xdist`	A string partially matching a survival time distribution to specify the external contact interval distribution. By default, it is the same as `dist`.
`init`	A named vector of initial values for estimated coefficients.
`fixed`	A named vector of fixed parameter values. These can include terms in `formula`, `intercept`, `logshape` for the internal shape parameter, and `xlogshape)` for the external shape parameter.
`optim_method`	The method to be used by `optim`.
`...`	Further arguments to be passed to `optim`.

Survival time distributions

The exponential distribution (exponential) has a rate parameter λ. Its hazard function is

h(t, λ) = λ.
The Weibull distribution (weibull) has a rate parameter λ and a shape parameter γ. Its cumulative hazard function is

H(t, λ, γ) = (λ t)^γ.

The exponential distribution is a Weibull distribution with shape γ = 1.
The log-logistic distribution (loglogistic) has a rate parameter lambda and a shape parameter γ. Its survival function is

S(t, λ, γ) = \frac{1}{1 + (λ t)^γ}.

Accelerated failure time models are log-linear models for the rate parameter λ in the specified survival time distribution. Each coefficient can be interpreted as the log rate ratio for a one-unit increase in the corresponding covariate, so each covariate has a multiplicative effect on the rate.

Internal and external transmission models The internal transmission model is for the hazard of transmission between individuals under observation. The external model is for hazard of transmission from external sources (e.g., the community or the environment) to individuals under observation. There are four types of covariates: internal-only, external-only, shared covariates with the same coefficients under both models, and shared covariates with (possibly) different coefficients under the two models. For covariates included in only one model, the covariate is set to zero for all data rows for the other model. Shared covariates with equal coefficients are reported as normal. Shared covariates that can have unequal coefficients are included as a shared main effect and an interaction with the external indicator variable. Thus, the main effect applies to the internal model and the main effect + interaction applies to the external model.

A list with class transreg that contains the following objects:

call: The call to transreg with complete formal arguments.
coefficients: Named vector of estimated parameters.
df: The number of estimated coefficients.
dist: String naming the internal contact interval distribution.
fixed: Named vector of fixed parameter values.
init: Named vector of initial parameter values.
loglik: The maximum log likelihood.
model_matrix: The data frame used to fit the model.
nlnL: Function for calculating the negative log likelihood. See details below.
optim_method: The method used in optim.
response: The response from Surv.
sus: Factor giving the susceptible member of each ordered pair.
var: The estimated variance matrix.
xdist: String naming the external contact interval distribution; NULL if model formula has no ext term.