在R语言中从文本字符串提取体温相关句子及数值

原始数据集 #

df <- data.frame(
  Person.ID = c(123L, 234L),
  Date = c("10/10/09", "11/11/03"),
  Text = c(
    "Here are some random words that I do not want. The person was allowed to cool to a core body temperature of 16.5 degrees centigrade. Here are some other random words I do not want.",
    "Here are some random words that I do not want. A cooling mechanism was applied to cool the patient to a core body temperature of 19.1 degrees centigrade. Here are some other random words I do not want."
  )
)

需求说明 #

为每个个体提取提及核心体温的句子（去除无关内容），同时新增仅包含体温数值的列，预期输出如下：

df2 <- data.frame(
  Person.ID = c(123L, 234L),
  Date = c("10/10/09", "11/11/03"),
  Text = c(
    "The person was allowed to cool to a core body temperature of 16.5 degrees centigrade.",
    "A cooling mechanism was applied to cool the patient to a core body temperature of 19.1 degrees centigrade. "
  ),
  Value = c(16.5, 19.1)
)

实现代码 #

用tidyverse工具集就能搞定，代码如下：

# 加载所需包
library(tidyverse)

# 处理数据
df2 <- df %>%
  # 提取包含核心体温的完整句子
  mutate(Text = str_extract(Text, "[^.]*core body temperature[^.]*\\.")) %>%
  # 提取体温数值并转为数值型
  mutate(Value = as.numeric(str_extract(Text, "\\d+\\.\\d+")))

代码解释 #

• str_extract(Text, "[^.]*core body temperature[^.]*\\.")：匹配从任意非句号字符开始、包含"core body temperature"、到下一个句号结束的内容，精准提取目标句子。
• str_extract(Text, "\\d+\\.\\d+")：匹配文本中的小数数值，再用as.numeric转换为数值类型，得到单独的体温数值列。

内容的提问来源于stack exchange，提问作者Jamie