Package 'RcmdrPlugin.survival'

Title: R Commander Plug-in for the 'survival' Package
Description: An R Commander plug-in for the survival package, with dialogs for Cox models, parametric survival regression models, estimation of survival curves, and testing for differences in survival curves, along with data-management facilities and a variety of tests, diagnostics and graphs.
Authors: John Fox
Maintainer: John Fox <[email protected]>
License: GPL (>= 2)
Version: 1.3-2
Built: 2024-11-18 06:26:23 UTC
Source: https://github.com/cran/RcmdrPlugin.survival

Help Index

Rcmdr Plug-In Package for the survival Package

Description

An R Commander plug-in for the survival package, with dialogs for managing survival data (this to a limited extent), Cox models, parametric survival regression models, estimation of survival curves, testing for differences in survival curves, and a variety of diagnostics, tests, and displays.

Details

The plug-in is tightly integrated with the R Commander interface; see the following menus: Data -> Survival data", Statistics -> Survival analysis, Statistics -> Fit Models, Models -> Hypothesis tests, Models -> Numerical diagnostics, Models -> Graphs.

Acknowledgments

I am grateful to Marilia Sa Carvalho, FIOCRUZ, Rio de Janeiro, Brazil, for many comments and suggestions, and to the following individuals for translations of messages into other languages: Philippe Grojean (French), Matjaz Jeran (Slovenian), Anton Korobeinikov (Russian), Manuel Munoz Marquez (Spanish), and Marilia Sa Carvalho (Portuguese).

Author(s)

John Fox

Maintainer: John Fox [email protected]

References

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32, doi:10.18637/jss.v049.i07.

Hemodialysis Data from Brazil

Description

This data set is analyzed in Sa Carvalho et al. (2003), and consists of data on 6805 hemodialysis patients in all federally funded clinics in Rio de Janeiro State, Brazil.

Usage

data(Dialysis)

Format

A data frame with 6805 observations on the following 7 variables.

center

a numeric code indicating in which of 67 centers the patient was treated.

age

of the patient.

begin

The month in which treatment began, with 1 representing January 1998.

end

The month in which observation terminated, either because of death or censoring. The study ended in month 44 (August, 2000).

event

1, death, or 0, censoring.

time

the difference between end and begin.

disease

a factor with levels congen, (congenital); diabetes; hypert (hypertension); other; and renal.

Source

M. Sa Carvalho, R. Henderson, S. Shimakura, and I. P. S. C. Sousa (2003). Survival of hemodialysis patients: Modeling differences in risk of dialysis centers. International Journal for Quality in Health Care, 15: 189–196.

References

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32. doi:10.18637/jss.v049.i07.

Examples

summary(Dialysis)
table(Dialysis$center)

Function to Compute Layout for Plot Array

Description

Given a number of plots n, find a arrangement for showing the plots in an array, set by par(mfrow=mfrow(n)).

Usage

mfrow(n, max.plots = 0)

Arguments

n

number of plots
max.plots

maximum number of plots; 0, the default, means no maximum.

Author(s)

John Fox <[email protected]>

See Also

par

Examples

mfrow(4)
mfrow(5)
mfrow(6)

Plot Method for coxph Objects

Description

Plots the predicted survival function from a coxph object, setting covariates to particular values.

Usage

## S3 method for class 'coxph'
plot(x, newdata, typical = mean,  byfactors=FALSE, 
  col = palette(), lty,  conf.level = 0.95, ...)

Arguments

x

a coxph object.
newdata

a data frame containing (combinations of) values to which predictors are set; optional.
typical

function to use to compute "typical" values of numeric predictors.
byfactors

if TRUE, different lines are drawn for each unique combination of factor values, including strata; if FALSE (the default) distinct lines are drawn only for different strata, with all columns of the model matrix (including for factors) set to their means.
col

colors for lines.
lty

line-types for lines; if missing, defaults to 1 to number required.
conf.level

level for confidence intervals; note: whether or not confidence intervals are plotted is determined by plot.survfit, which plot.coxph calls; if a conf.int argument is supplied it is passed through.
...

arguments passed to plot.

Details

If newdata is missing then all combinations of levels of factor-predictors (or strata), if present, are combined with "typical" values of numeric predictors.

Value

Invisibly returns the summary resulting from applying survfit.coxph to the coxph object.

Author(s)

John Fox [email protected].

References

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32. doi:10.18637/jss.v049.i07.

See Also

coxph, survfit.coxph, plot.survfit.

Examples

require(survival)
cancer$sex <- factor(ifelse(cancer$sex == 1, "male", "female"))

mod.1 <- coxph(Surv(time, status) ~ age + wt.loss, data=cancer)
plot(mod.1)
plot(mod.1, typical=function(x) quantile(x, c(.25, .75)))

mod.2 <- coxph(Surv(time, status) ~ age + wt.loss + sex, data=cancer)
plot(mod.2)

mod.3 <- coxph(Surv(time, status) ~ (age + wt.loss)*sex, data=cancer)
plot(mod.3)

mod.4 <- coxph(Surv(time, status) ~ age + wt.loss + strata(sex), data=cancer)
plot(mod.4)

mods.1 <- survreg(Surv(time, status) ~ age + wt.loss, data=cancer)

Rossi et al.'s Criminal Recidivism Data

Description

This data set is originally from Rossi et al. (1980), and is used as an example in Allison (1995). The data pertain to 432 convicts who were released from Maryland state prisons in the 1970s and who were followed up for one year after release. Half the released convicts were assigned at random to an experimental treatment in which they were given financial aid; half did not receive aid.

Usage

Rossi

Format

A data frame with 432 observations on the following 62 variables.

week

week of first arrest after release or censoring; all censored observations are censored at 52 weeks.

arrest

1 if arrested, 0 if not arrested.

fin

financial aid: no yes.

age

in years at time of release.

race

black or other.

wexp

full-time work experience before incarceration: no or yes.

mar

marital status at time of release: married or not married.

paro

released on parole? no or yes.

prio

number of convictions prior to current incarceration.

educ

level of education: 2 = 6th grade or less; 3 = 7th to 9th grade; 4 = 10th to 11th grade; 5 = 12th grade; 6 = some college.

emp1

employment status in the first week after release: no or yes.

emp2

as above.

emp3

as above.

emp4

as above.

emp5

as above.

emp6

as above.

emp7

as above.

emp8

as above.

emp9

as above.

emp10

as above.

emp11

as above.

emp12

as above.

emp13

as above.

emp14

as above.

emp15

as above.

emp16

as above.

emp17

as above.

emp18

as above.

emp19

as above.

emp20

as above.

emp21

as above.

emp22

as above.

emp23

as above.

emp24

as above.

emp25

as above.

emp26

as above.

emp27

as above.

emp28

as above.

emp29

as above.

emp30

as above.

emp31

as above.

emp32

as above.

emp33

as above.

emp34

as above.

emp35

as above.

emp36

as above.

emp37

as above.

emp38

as above.

emp39

as above.

emp40

as above.

emp41

as above.

emp42

as above.

emp43

as above.

emp44

as above.

emp45

as above.

emp46

as above.

emp47

as above.

emp48

as above.

emp49

as above.

emp50

as above.

emp51

as above.

emp52

as above.

Source

Allison, P.D. (1995). Survival Analysis Using the SAS System: A Practical Guide. Cary, NC: SAS Institute.

References

Rossi, P.H., R.A. Berk, and K.J. Lenihan (1980). Money, Work, and Crime: Some Experimental Results. New York: Academic Press.

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32. doi:10.18637/jss.v049.i07.

Examples

summary(Rossi)

Define Survival Data Dialog Box

Description

This dialog box permits you to define a time variable (or start and stop variables), an event indicator, a strata variable or variables, and a cluster variable to be associated with the current data set. If these characteristics are defined, then they will become default choices where appropriate in other dialog boxes.

Usage

SurvivalData() # normally not called directly

Value

Used only for its side effect.

Author(s)

John Fox <[email protected]>

References

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32. doi:10.18637/jss.v049.i07.

Diagnostics for Survival Regression Models

Description

These are primarily convenience functions for the RcmdrPlugin.survival package, to produce diagnostics for coxph and survreg models in a convenient form for plotting via the package's GUI.

Usage

crPlots(model, ...)
## S3 method for class 'coxph'
crPlots(model, ...)

## S3 method for class 'coxph'
dfbeta(model, ...)
## S3 method for class 'dfbeta.coxph'
plot(x, ...)

## S3 method for class 'coxph'
dfbetas(model, ...)
## S3 method for class 'dfbetas.coxph'
plot(x, ...)

## S3 method for class 'survreg'
dfbeta(model, ...)
## S3 method for class 'dfbeta.survreg'
plot(x, ...)

## S3 method for class 'survreg'
dfbetas(model, ...)
## S3 method for class 'dfbetas.survreg'
plot(x, ...)

MartingalePlots(model, ...)
## S3 method for class 'coxph'
MartingalePlots(model, ...)

testPropHazards(model, test.terms = FALSE, plot.terms = FALSE, ...)
## S3 method for class 'coxph'
testPropHazards(model, test.terms = FALSE, plot.terms = FALSE, ...)

Arguments

model, x

a Cox regression or parametric survival regression model, as appropriate.
test.terms

test proportional hazards by terms in the Cox model, rather than by coefficients (default is FALSE).
plot.terms

diagnostic plots of proportional hazards by terms in the Cox model, rather than by coefficients (default is FALSE).
...

arguments to be passed down.

Details

  • crPlots.coxph is a method for the crPlots function in the car package, to create component+residual (partial-residual) plots, using residuals.coxph and predict.coxph in the survival package.

  • testPropHazards is essentially a wrapper for the cox.zph function in the survival package.

  • MartingalePlots creates null-model Martingale plots for Cox regression models, using the residuals.coxph function in the survival package.

  • dfbeta.coxph and dfbetas.coxph provide methods for the standard dfbeta and dfbetas functions, using the residuals.coxph function in the survival package for computation. plot.dfbeta.coxph and plot.dfbetas.coxph are plot methods for the objects produced by these functions.

  • dfbeta.survreg, dfbetas.survreg, plot.dfbeta.survreg and plot.dfbetas.survreg are similar methods for survreg objects.

Value

Most of these function create graphs and don't return useful values; the dfbeta and dfbetas methods create matrices of dfbeta and dfbetas values.

Author(s)

John Fox <[email protected]>

References

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32. doi:10.18637/jss.v049.i07.

See Also

coxph, survreg, crPlots, residuals.coxph, residuals.survreg, predict.coxph, cox.zph

Convert a Survival Data Set from "Wide" to "Long" Format

Description

Converts a survival-analysis data frame from "wide" format, in which time-varying covariates are separate variables, one per occasion, to "long" or counting-process format in which each occasion is a separate row in the data frame.

Usage

unfold(data, ...)

## S3 method for class 'data.frame'
unfold(data, time, event, cov, 
  cov.names = paste("covariate", ".", 1:ncovs, sep = ""), 
  suffix = ".time", cov.times = 0:ncov, common.times = TRUE, lag = 0, 
  show.progress=TRUE, ...)

Arguments

data

a data frame to be "unfolded" from wide to long.
time

the column number or quoted name of the event/censoring-time variable in data.
event

the column number or quoted name of the event/censoring-indicator variable in data.
cov

a vector giving the column numbers of the time-dependent covariate in data, or a list of vectors if there is more than one time-varying covariate.
cov.names

a character string or character vector giving the name or names to be assigned to the time-dependent covariate(s) in the output data set.
suffix

the suffix to be attached to the name of the time-to-event variable in the output data set; defaults to '.time'.
cov.times

the observation times for the covariate values, including the start time. This argument can take several forms: (1) The default is integers from 0 to the number of covariate values (i.e., one more than the length of each vector in cov). (2) An arbitrary numerical vector with one more entry than the length of each vector in cov. (3) The columns in the input data set that give the observations times for each individual. There should be one more column than the length of each vector in cov.
common.times

a logical value indicating whether the times of observation are the same for all individuals; defaults to TRUE.
lag

number of observation periods to lag each value of the time-varying covariate(s); defaults to 0.
show.progress

if TRUE, the default, show a progress bar as the observations are processed.
...

arguments to be passed down.

Value

A data frame containing the "long" version of the data set.

Author(s)

John Fox <[email protected]>

References

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32. doi:10.18637/jss.v049.i07.

Examples

if (interactive()){
	head(Rossi, 2)
	Rossi.long <- unfold(Rossi, time="week", event="arrest", cov=11:62, 
	  cov.names="emp")
	head(Rossi.long, 50)
}

Dialog to Convert a Survival Data Set from "Wide" to "Long" Format

Description

Converts a survival-analysis data frame from "wide" format, in which time-varying covariates are separate variables, one per occasion, to "long" or counting-process format in which each occasion is a separate row in the data frame.

Usage

Unfold() # called via the R Commander menus

Details

Most of the dialog box is self-explanatory. A time-varying covariate is identified by selecting the variables constituting the covariate in the "wide" version of the data set using the variable-list box at the lower-left; specifying a name to be used for the covariate in the "long" version of the data set; and pressing the Select button. This process is repeated for each time-varying covariate. All time-varying covariates have to be measured on the same occasions, which are assigned times 0, 1, ... in the output data set. If the covariates are to be lagged, this is indicated via the Lag covariates slider near the lower right. The default lag is 0 — i.e., no lag. The output data set will include variables named start and stop, which give the counting-process start and stop times for each row, and an event indicator composed of the name of the event indicator in the "wide" form of the data set and the suffix .time.

The Unfold dialog calls the unfold function, which is somewhat more flexible.

Author(s)

John Fox <[email protected]>

References

John Fox, Marilia Sa Carvalho (2012). The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis. Journal of Statistical Software, 49(7), 1-32. doi:10.18637/jss.v049.i07.

See Also

unfold