The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

jskm

Jinseob Kim

2024-07-08

Install

install.packages("devtools")
library(devtools)
install_github("jinseob2kim/jskm")
library(jskm)

Example

Survival probability

# Load dataset
library(survival)
data(colon)
#> Warning in data(colon): data set 'colon' not found
fit <- survfit(Surv(time, status) ~ rx, data = colon)

# Plot the data
jskm(fit)

jskm(fit,
  table = T, pval = T, label.nrisk = "No. at risk", size.label.nrisk = 8,
  xlabs = "Time(Day)", ylabs = "Survival", ystratalabs = c("Obs", "Lev", "Lev + 5FU"), ystrataname = "rx",
  marks = F, timeby = 365, xlims = c(0, 3000), ylims = c(0.25, 1), showpercent = T
)
#> Warning: Removed 16 rows containing missing values or values outside the scale range
#> (`geom_step()`).
#> Warning: Removed 3 rows containing missing values or values outside the scale range
#> (`geom_text()`).

Cumulative incidence: 1- Survival probability

jskm(fit, ci = T, cumhaz = T, mark = F, ylab = "Cumulative incidence (%)", surv.scale = "percent", pval = T, pval.size = 6, pval.coord = c(300, 0.7))

Landmark analysis

jskm(fit, mark = F, surv.scale = "percent", pval = T, table = T, cut.landmark = 500, showpercent = T)

Competing risk analysis

status2 variable: 0 - censoring, 1 - event, 2 - competing risk

## Make competing risk variable, Not real
colon$status2 <- colon$status
colon$status2[1:400] <- 2
colon$status2 <- factor(colon$status2)
fit2 <- survfit(Surv(time, status2) ~ rx, data = colon)
jskm(fit2, mark = F, surv.scale = "percent", table = T, status.cmprsk = "1")

jskm(fit2, mark = F, surv.scale = "percent", table = T, status.cmprsk = "1", showpercent = T, cut.landmark = 500)

Theme Jama

jskm(fit, theme = "jama", cumhaz = T, table = T, mark = F, ylab = "Cumulative incidence (%)", surv.scale = "percent", pval = T, pval.size = 6, pval.coord = c(300, 0.7))

Theme Nejmoa

jskm(fit, theme = "nejm", nejm.infigure.ratiow = 0.6, nejm.infigure.ratioh = 0.4, nejm.infigure.ylim = c(0, 0.7), cumhaz = T, table = T, mark = F, ylab = "Cumulative incidence (%)", surv.scale = "percent", pval = T, pval.size = 6, pval.coord = c(300, 0.7))

Weighted Kaplan-Meier plot - svykm.object in survey package

library(survey)
#> Loading required package: grid
#> Loading required package: Matrix
#> 
#> Attaching package: 'survey'
#> The following object is masked from 'package:graphics':
#> 
#>     dotchart
data(pbc, package = "survival")
pbc$randomized <- with(pbc, !is.na(trt) & trt > 0)
biasmodel <- glm(randomized ~ age * edema, data = pbc)
pbc$randprob <- fitted(biasmodel)

dpbc <- svydesign(id = ~1, prob = ~randprob, strata = ~edema, data = subset(pbc, randomized))

s1 <- svykm(Surv(time, status > 0) ~ 1, design = dpbc)
s2 <- svykm(Surv(time, status > 0) ~ sex, design = dpbc)

svyjskm(s1)

svyjskm(s2)

svyjskm(s2, cumhaz = T, ylab = "Cumulative incidence(%)", surv.scale = "percent", showpercent = T)

If you want to get confidence interval, you should apply se = T option to svykm object.

s3 <- svykm(Surv(time, status > 0) ~ sex, design = dpbc, se = T)
svyjskm(s3)

svyjskm(s3, ci = F, showpercent = T)

svyjskm(s3, ci = F, surv.scale = "percent", pval = T, table = T, cut.landmark = 1000)

Theme

JAMA

svyjskm(s2, theme = "jama", pval = T, table = T, design = dpbc)

NEJM

svyjskm(s2, theme = "nejm", nejm.infigure.ratiow = 0.4, nejm.infigure.ratioh = 0.4, nejm.infigure.ylim = c(0.2, 1), pval = T, table = T, design = dpbc)

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.