了解 coxph 模型的 R survfit.plot 函数中的颜色答案

【问题标题】：Understand colors in R survfit.plot function for a coxph model了解 coxph 模型的 R survfit.plot 函数中的颜色
【发布时间】：2021-08-10 08:30:37
【问题描述】：

我正在尝试创建一组带有分类解释变量的单变量 coxph 模型，对于每个模型，我想按变量级别比较生存函数。例如我正在创建：

library(survival)

set.seed(4321)
data <- data.frame(cbind(
  sample(0:1, 40, replace=TRUE), # event
  sample(0:24, 40, replace=TRUE), # time
  sample(1:2, 40, replace=TRUE), # Diam
  sample(3:4, 40, replace=TRUE) # N
  ))

colnames(data) <- c("event","time","Diam","N")

data$Diam <- as.factor(data$Diam)
data$N <- as.factor(data$N)

model1 <- coxph(Surv(time, event) ~ Diam, data = data)
model2 <- coxph(Surv(time, event) ~ N, data = data)

其中 Diam 和 N 有两个级别。我准备了这段代码来绘制生存函数：

plot(survfit(model1, newdata = data), fun = "s",
     conf.int = TRUE,
     col = 2:4,
     xlab = "Months",
     ylab = "Proportion without event",
     main = "Survival curve by Diam")
legend("bottomleft", 
       legend = levels(data$Diam),
       lty = 1, col = c(2,4))

plot(survfit(model2, newdata = data), fun = "s",
     conf.int = TRUE,
     col = 2:4,
     xlab = "Months",
     ylab = "Proportion without event",
     main = "Survival curve by N")
legend("bottomleft", 
       legend = levels(data$N),
       lty = 1, col = c(2,4))

这是输出：

我不明白：

为什么第二张图表的颜色不同？
如何匹配图表和图例中的颜色。

任何帮助将不胜感激！先感谢您，弗朗西斯卡

【问题讨论】：

如果您包含一个简单的reproducible example 以及可用于测试和验证可能的解决方案的示例输入，则更容易为您提供帮助。
@MrFlick：谢谢你的建议，我已经完成了。希望对您有所帮助。

标签： r plot survival-analysis cox-regression

【解决方案1】：

我觉得很奇怪，你在绘制情节时使用了三种颜色，而在图例中只使用了两种颜色。我设法用这个正确地绘制它

plot(survfit(model1, newdata = data), fun = "s",
     conf.int = TRUE,
     col = c(1,2,2),
     xlab = "Months",
     ylab = "Proportion without event",
     main = "Survival curve by Diam")
legend("bottomleft", 
       legend = levels(data$Diam),
       lty = 1, col = levels(data$Diam))

plot(survfit(model2, newdata = data), fun = "s",
     conf.int = TRUE,
     col = levels(data$N),
     xlab = "Months",
     ylab = "Proportion without event",
     main = "Survival curve by N")
legend("bottomleft", 
       legend = levels(data$N),
       lty = 1, col = levels(data$N))

但请注意，我在第一个情节中保留了三种颜色，而在第二个情节中只保留了两种。我认为问题出在模型或采样上。

此外，使用不同的种子（例如 1），颜色不会有任何问题。

set.seed(1)
data <- data.frame(cbind(
  sample(0:1, 40, replace=TRUE), # event
  sample(0:24, 40, replace=TRUE), # time
  sample(1:2, 40, replace=TRUE), # Diam
  sample(3:4, 40, replace=TRUE) # N
))

colnames(data) <- c("event","time","Diam","N")

data$Diam <- as.factor(data$Diam)
data$N <- as.factor(data$N)

model1 <- coxph(Surv(time, event) ~ Diam, data = data)
model2 <- coxph(Surv(time, event) ~ N, data = data)
plot(survfit(model1, newdata = data), fun = "s",
     conf.int = TRUE,
     col = levels(data$Diam),
     xlab = "Months",
     ylab = "Proportion without event",
     main = "Survival curve by Diam")
legend("bottomleft", 
       legend = levels(data$Diam),
       lty = 1, col = levels(data$Diam))

plot(survfit(model2, newdata = data), fun = "s",
     conf.int = TRUE,
     col = levels(data$N),
     xlab = "Months",
     ylab = "Proportion without event",
     main = "Survival curve by N")
legend("bottomleft", 
       legend = levels(data$N),
       lty = 1, col = levels(data$N))

干杯，

【讨论】：