ggsurvplot 将观察数解释为 newdata 参数中的级别

Question

我遇到了 R 中 survminer 包中的 ggsurvplot 函数的问题。当尝试使用带有 newdata 参数的 ggsurvplot 绘制生存曲线时，该函数错误地将观察数解释为级别数，从而导致错误。如果没有 newdata 参数，该函数将无法正确识别因子变量的级别，从而导致另一个错误。

观察到的行为： 使用 newdata 参数：

set.seed(123) 
d.dat.tot <- data.frame( 
  DONOR_SURVIVAL_TIME = rexp(100, 0.1), 
  DEAD_OR_ALIVE = sample(0:1, 100, replace = TRUE), 
  GENOTYPE.111758446 = sample(c("A/A", "G/A", "G/G"), 100, replace = TRUE) 
  ) 
# Data preparation 
d.dat.tot.clean <- d.dat.tot[d.dat.tot$DEAD_OR_ALIVE < 2, ] 
d.dat.tot.clean$GENOTYPE.111758446 <- as.factor(d.dat.tot.clean$GENOTYPE.111758446) 

cox_model_single_111758446 <- coxph(
  Surv(DONOR_SURVIVAL_TIME, DEAD_OR_ALIVE) ~ GENOTYPE.111758446, 
  data = d.dat.tot.clean
)

cox_plot_single_111758446 <- ggsurvplot(
  survfit(cox_model_single_111758446, newdata = d.dat.tot.clean),
  data = d.dat.tot.clean,
  pval = TRUE,
  conf.int = TRUE,
  risk.table = TRUE,
  legend.title = "Genotype",
  legend.labs = levels(d.dat.tot.clean$GENOTYPE.111758446),
  xlab = "Time to Last Follow-up, mo",
  ylab = "Cumulative Survival, %",
  ggtheme = theme_minimal(),
  palette = "set2"
)

这会导致错误：

Error in ggsurvplot_df(d, fun = fun, color = color, palette = palette, :
The length of legend.labs should be 236.

（236是我的案例数。）

没有 newdata 参数： 该函数无法正确识别因子变量的级别并返回：

Error in ggsurvplot_df(d, fun = fun, color = color, palette = palette, :
The length of legend.labs should be 1.

# Sample reproducible data
set.seed(123)
d.dat.tot <- data.frame(
  DONOR_SURVIVAL_TIME = rexp(100, 0.1),
  DEAD_OR_ALIVE = sample(0:1, 100, replace = TRUE),
  GENOTYPE.111758446 = sample(c("A/A", "G/A", "G/G"), 100, replace = TRUE)
)

# Data preparation
d.dat.tot.clean <- d.dat.tot[d.dat.tot$DEAD_OR_ALIVE < 2, ]
d.dat.tot.clean$GENOTYPE.111758446 <- as.factor(d.dat.tot.clean$GENOTYPE.111758446)

# Cox model fitting
cox_model_single_111758446 <- coxph(Surv(DONOR_SURVIVAL_TIME, DEAD_OR_ALIVE) ~ GENOTYPE.111758446, data = d.dat.tot.clean)

# Survival curves creation
surv_fit <- survfit(cox_model_single_111758446)

# Plotting survival curves
cox_plot_single_111758446 <- ggsurvplot(
  surv_fit,
  data = d.dat.tot.clean,
  pval = TRUE,
  conf.int = TRUE,
  risk.table = TRUE,
  legend.title = "Genotype",
  legend.labs = levels(d.dat.tot.clean$GENOTYPE.111758446),
  xlab = "Time to Last Follow-up, mo",
  ylab = "Cumulative Survival, %",
  ggtheme = theme_minimal(),
  palette = "set2"
)

print(cox_plot_single_111758446)

Answer 1

我没有完整的答案。我认为您需要为每组提供 1 行

newdata

，这适用于绘图，但不适用于表格。

sfit <- survfit(
  cox_model_single_111758446,
  newdata = data.frame(GENOTYPE.111758446 = levels(d.dat.tot.clean$GENOTYPE.111758446))
)

cox_plot_single_111758446 <- ggsurvplot(
  sfit,
  data = d.dat.tot.clean,
  pval = TRUE,
  conf.int = TRUE,
  risk.table = TRUE,
  legend.title = "Genotype",
  legend.labs = levels(d.dat.tot.clean$GENOTYPE.111758446),
  xlab = "Time to Last Follow-up, mo",
  ylab = "Cumulative Survival, %",
  ggtheme = theme_minimal(),
  palette = "set2"
)