Synthetic population data

Description

This is made-up time-to-event data with properties that make it useful for testing and demonstrating svycoxme functions. There is a single level of clustering, identified with group_id, and the X covariates depend on Z covariates.

Usage

pop

Format

A data frame with 20,000 rows and 10 columns:

X1

Observation-level N(\mu_{X1}, 1) distributed covariate where \mu_{X1}=0.5*(Z1 + 1)

X2

Cluster-level N(\mu_{X2}, 1) distributed covariate where \mu_{X2}=0.5*(Z2 + Z3)

X3

Cluster-level binary covariate where Pr(X3=1) = Z3

Z1

Stratum membership. Takes the values 1 to 5

Z2

cluster-level N(0,1) distributed covariate

Z3

cluster-level Uniform(0,1) distributed covariate

stat_time

Event or Censoring time

stat

Event/Censoring indicator. Event=1; Censoring=0

group_id

Unique cluster ID

obs_id

Unique observation ID

sampled

Sampling indicator. Is this observation in samp_srcs?

Calculate residuals for a 'coxme' fit

Description

Calculates score, dfbeta, or dfbetas residuals for a mixed-effects proportional hazards model. Only fixed-effect components are calculated; see Details.

Usage

## S3 method for class 'coxme'
residuals(
  object,
  data,
  type = c("score", "dfbeta", "dfbetas"),
  weighted = (type %in% c("dfbeta", "dfbetas")),
  include_re = FALSE,
  ...
)

Arguments

Details

An observation's contribution to the score vector includes values for every fixed and random effect in the fitted model. In many cases, the number of random effects will be large, and most residuals will be zero. Until efficient sparse computation is implemented, it is too expensive computationally and on memory to calculate the random effect residual terms, so they are excluded. This is likely to change, and the parameter include_re is include for future expansion.

Value

A matrix of residuals. The score residuals are each observation's contribution to the score vector. Two transformations of this are often more useful: dfbeta is the approximate change in the coefficient vector if that observation were dropped, and dfbetas is the approximate change in the coefficients, scaled by the standard error for the coefficients.

Examples

fit1 <- coxme::coxme(survival::Surv(stat_time, stat) ~ X1 + X2 + X3 + (1 | group_id),
                     data = samp_srcs)
dfbeta_res <- resid(fit1, data = samp_srcs, type = "dfbeta")

head(dfbeta_res)

Simple random cluster sample of 100 clusters from synthetic population data, pop.

Description

This is made-up time-to-event data with properties that make it useful for testing and demonstrating svycoxme functions. There is a single level of clustering, identified with group_id, and the X covariates depend on Z covariates.

Usage

samp_srcs

Format

A data frame with 20,000 rows and 10 columns:

X1

Observation-level N(\mu_{X1}, 1) distributed covariate where \mu_{X1}=0.5*(Z1 + 1).

X2

Cluster-level N(\mu_{X2}, 1) distributed covariate where \mu_{X2}=0.5*(Z2 + Z3).

X3

Cluster-level binary covariate where Pr(X3=1) = Z3.

Z1

Stratum membership. Takes the values 1 to 5.

Z2

cluster-level N(0,1) distributed covariate.

Z3

cluster-level Uniform(0,1) distributed covariate.

stat_time

Event or Censoring time.

stat

Event/Censoring indicator. Event=1; Censoring=0.

group_id

Unique cluster ID.

obs_id

Unique observation ID.

fpc

Total number of clusters in the population.

weight

Observation-level inverse probability of selection weight.

Survey-weighted mixed-effects Cox models

Description

Fit a mixed-effect proportional hazards model to data from a complex design.

Usage

svycoxme(
  formula,
  design,
  subset = NULL,
  rescale = TRUE,
  control = coxme::coxme.control(),
  ...
)

## S3 method for class 'DBIsvydesign'
svycoxme(
  formula,
  design,
  subset = NULL,
  rescale = TRUE,
  control = coxme::coxme.control(),
  ...
)

## S3 method for class 'survey.design'
svycoxme(
  formula,
  design,
  subset = NULL,
  rescale = TRUE,
  control = coxme::coxme.control(),
  ...
)

## S3 method for class 'svyrep.design'
svycoxme(
  formula,
  design,
  subset = NULL,
  rescale = TRUE,
  control = coxme::coxme.control(),
  multicore = getOption("survey.multicore"),
  return.replicates = FALSE,
  ...
)

Arguments

Details

Parallel processing is done with future_lapply. Future planning is left to the user, e.g. using plan before the call to svycoxme. Note that svycoxme.DBIsvydesign has not been implemented yet.

Value

An object of class svycoxme.

Examples

des <- survey::svydesign(ids = ~group_id, weights = ~weight, data = samp_srcs)
fit1 <- svycoxme(survival::Surv(stat_time, stat) ~ X1 + X2 + X3 + (1 | group_id),
                 design = des)
summary(fit1)

# with replicate weights (only 10 replicates are used to reduce CPU time)
repdes <- survey::as.svrepdesign(des, type = "bootstrap", replicates = 10)
fit2 <- svycoxme(survival::Surv(stat_time, stat) ~ X1 + X2 + X3 + (1 | group_id),
                 design = repdes)
summary(fit2)

# use multicore processing (`n_cores = 2` to comply with CRAN policy). Otherwise,
# something like, `floor(parallelly::availableCores() * 0.8)`, could be used.

n_cores = 2
future::plan("multicore", workers = n_cores)
fit3 <- svycoxme(survival::Surv(stat_time, stat) ~ X1 + X2 + X3 + (1 | group_id),
                 design = repdes, multicore = TRUE)
all.equal(coef(fit2), coef(fit3))
future::plan("sequential")