如何在R中将诊断编码分组为疾病类别新变量

方法1：修正case_when()的用法 #

你之前的代码错误在于用==匹配向量（会引发循环匹配问题），且不需要嵌套case_when。正确做法是用%in%判断编码是否属于目标组，代码如下：

library(dplyr)

data <- data %>%
  mutate(diagnosis = case_when(
    diag_code %in% c(1,2,3,4,84) ~ "resp disease",
    diag_code %in% c(5,6,7,32,56) ~ "CVD disease",
    TRUE ~ "Unknown"
  ))

• %in%是匹配多个值的正确方式，用来检查diag_code是否在指定编码列表中
• case_when按顺序匹配条件，第一个符合的条件会返回对应类别
• 最后一行TRUE ~ "Unknown"处理所有未匹配到的编码

方法2：用recode()快速映射 #

如果编码和类别对应关系简单，recode()是更简洁的选择：

data <- data %>%
  mutate(diagnosis = recode(
    diag_code,
    `1` = "resp disease", `2` = "resp disease", `3` = "resp disease",
    `4` = "resp disease", `84` = "resp disease",
    `5` = "CVD disease", `6` = "CVD disease", `7` = "CVD disease",
    `32` = "CVD disease", `56` = "CVD disease",
    .default = "Unknown"
  ))

若需要diagnosis为因子类型，替换成recode_factor()即可。

方法3：映射表合并（适合大量编码场景） #

如果后续要频繁修改或扩展疾病类别，建议先创建映射表，再通过合并实现分组，这种方式更易维护：

# 创建诊断编码-疾病类别的映射表
diag_map <- tibble(
  diag_code = c(1,2,3,4,84,5,6,7,32,56),
  diagnosis = c(rep("resp disease",5), rep("CVD disease",5))
)

# 合并原数据和映射表，处理未匹配的编码
data <- data %>%
  left_join(diag_map, by = "diag_code") %>%
  mutate(diagnosis = ifelse(is.na(diagnosis), "Unknown", diagnosis))

内容的提问来源于stack exchange，提问作者Alaina