以SOFA 评分表为例：
image.png
- 模拟数据
pao2 <- round(rnorm(100,mean = 300,sd = 30))
summary(pao2)
fio2 <- round(rnorm(100,mean = 0.5,sd = 0.1),2)
gcs <- round(rnorm(100,mean = 80,sd = 20)/10)
map <- round(rnorm(100,mean = 65,sd = 15))
dop <- round(rnorm(100,mean = 10,sd = 3),1)
dob <- rbinom(100,1,0.4)
epi <- round(rnorm(100,mean = 10,sd = 3)/100,2)
nor <- round(rnorm(100,mean = 10,sd = 3)/100,2)
bilirubin <- round(rnorm(100,mean = 80,sd = 20),1)
platelet <- round(rnorm(100,mean = 180,sd =50))
cr <- round(rnorm(100,mean = 150,sd = 34))
uo <- rnorm(100, mean = 1000, sd = 500)
uo <- round(ifelse(uo >0, uo,-uo)) # 使其不出现负数
data <- data.frame(pao2,fio2,gcs,map,dop,dob,epi,nor,bilirubin,platelet,cr,uo)
head(data,4)
# pao2 fio2 gcs map dop dob epi nor bilirubin platelet cr uo
# 1 306 0.55 10 60 9.1 0 0.13 0.07 57.8 152 113 2158
# 2 306 0.62 10 75 12.5 1 0.14 0.14 53.5 237 126 436
# 3 274 0.51 9 91 9.8 1 0.11 0.07 68.0 135 104 868
# 4 312 0.38 8 63 7.8 1 0.07 0.11 78.2 190 133 1230
- 将呼吸、凝血、肝功指标转化为相应评分
data$respiratory <- ifelse(data$pao2/data$fio2 >= 400,0,
ifelse(data$pao2/data$fio2 >= 300,1,
ifelse(data$pao2/data$fio2 >= 200,2,
ifelse(data$pao2/data$fio2 >= 100,3,4))))
data$coagulation <- ifelse(data$platelet >= 150,0,
ifelse(data$platelet >= 100,1,
ifelse(data$platelet >= 50,2,
ifelse(data$platelet >= 20,3,4))))
data$liver <- ifelse(data$bilirubin >= 204,4,
ifelse(data$bilirubin >= 102,3,
ifelse(data$bilirubin >= 33,2,
ifelse(data$bilirubin >= 20,1,0))))
head(data[,c("pao2","fio2","respiratory","coagulation","liver")])
# pao2 fio2 respiratory coagulation liver
# 1 306 0.55 0 0 2
# 2 306 0.62 0 0 2
# 3 274 0.51 0 1 2
# 4 312 0.38 0 0 2
# 5 348 0.52 0 1 2
# 6 271 0.56 0 0 3
- 转化神经系统、肾功、心功指标为相应评分
# 中枢
data$neuro <- cut(data$gcs,breaks = c(3,5,9,12,14,15),labels = c(4,3,2,1,0),include.lowest = TRUE)
data$neuro <- as.numeric(as.character(data$neuro)) #将上一步的结果为factor
data$neuro
# 肾功
cr.score <- ifelse(data$cr < 110,0,
ifelse(data$cr <= 170,1,
ifelse(data$cr <= 229,2,
ifelse(data$cr <= 440,3,4))))
uo.score <- ifelse(data$uo >= 500,0,ifelse(data$uo >200,3,4))
data$renal <- max(cr.score,uo.score)
# 循环
map.score <- ifelse(data$map >= 70,0,1)
dop.score <- ifelse(data$dop > 15,4,ifelse(data$dop >5.1,3,2))
dob.score <- ifelse(data$dob == 1,2,0)
epi.score <- ifelse(data$epi <= 0.1,3,4)
nor.score <- ifelse(data$nor > 0.1,4,3)
data$cardio <- max(map.score,dop.score,dob.score,epi.score,nor.score)
# sofa 评分
data$sofa.score <- rowSums(data[,c("cardio","renal","coagulation","liver","neuro","respiratory")])
head(data)
image.png
- 连续变量重新编码
data$oxyindex <- data$pao2/data$fio2
data$berlin <- cut(data$oxyindex, breaks = c(min(data$oxyindex),
100,200,300,max(data$oxyindex)),
labels = c("severe","moderate","mild","none"),include.lowest = TRUE)
table(data$berlin)
# severe moderate mild none
# 0 0 1 99
- 修改变量名
# method 1
names(data)[20] <- "oxygen.index"
names(data)[11] <- "creatinine"
names(data)
# method 2
library(reshape)
data <- rename(data,c(gcs = "GCS",berlin = "Berlin"))
names(data)
参考资料
文中代码来自章仲恒教授的丁香园课程：R语言基本操作
