在回顾前几行的同时创建一个新变量

Question

我有一个简单的患者就诊数据集：

date        infection
2005-01-01  yes
2005-06-30  yes
2005-10-15  yes
2006-01-01  no
2006-06-01  no
2006-11-01  yes
2006-12-01  no
2007-11-15  yes

在 R 中，我想添加一个名为 chronic 的列，它的值是 yes、no、NA.

仅当感染=='yes' 当前日期并且在过去 365 天内有两个感染=='yes' 行时才显示值 yes。
否则，如果在过去的 365 天内没有两次访问，则需要 NA
否则需要no

所以最终的数据集看起来像这样：

date        infection chronic
2005-01-01  yes       NA
2005-06-30  yes       NA
2005-10-15  yes       yes
2006-01-01  no        no
2006-06-01  no        no
2006-11-01  yes       no
2006-12-01  no        no
2007-11-15  yes       NA

我该如何编码？理想情况下，我想使用 dplyr，但我愿意接受任何解决方案。谢谢！

可以使用此代码重新创建数据集：

dat <- data.frame(date = c(as.Date("2005-01-01"), as.Date("2005-06-30"), as.Date("2005-10-15"), as.Date("2006-01-01"), as.Date("2006-06-01"), as.Date("2006-11-01"), as.Date("2006-12-01"), as.Date("2007-11-15")), infection = c("yes", "yes", "yes", "no", "no", "yes", "no", "yes"))

Answer 1

您可以尝试使用 purrr 中的 map 函数：

library(dplyr)
library(purrr)

dat %>%
  mutate(chronic = map2_chr(date, infection, 
         ~case_when(.y == 'yes' & 
                    sum(infection[between(date, .x-365, .x - 1)] == 'yes') >= 2 ~ 'yes', 
                    .y == 'yes' & 
                    sum(infection[between(date, .x-365, .x - 1)] == 'yes') != 2 ~ NA_character_, 
                    TRUE ~ 'no')))

#        date infection chronic
#1 2005-01-01       yes    <NA>
#2 2005-06-30       yes    <NA>
#3 2005-10-15       yes     yes
#4 2006-01-01        no      no
#5 2006-06-01        no      no
#6 2006-11-01       yes    <NA>
#7 2006-12-01        no      no
#8 2007-11-15       yes    <NA>

Answer 2

使用 data.table 和基于范围（非等值）的替代方案 merge/join。

library(data.table)
library(magrittr) # not required, just used to show the flow

dat <- fread(text = "
date        infection
2005-01-01  yes
2005-06-30  yes
2005-10-15  yes
2006-01-01  no
2006-06-01  no
2006-11-01  yes
2006-12-01  no
2007-11-15  yes")[, date := as.Date(date)]

代码：

copy(dat) %>%
  .[, c("date0", "date1") := .(date - 365, date)] %>%
  dat[., on = .(date >= date0, date <= date1) ] %>%
  .[, .(infection = last(infection), n_visits = .N,
        n_infect = sum(infection == "yes")), by = .(i.date)] %>%
  setnames(., "i.date", "date") %>%
  .[, chronic := fcase(
    n_visits < 3, NA_character_,
    infection == "yes" & n_infect >= 2, "yes",
    rep(TRUE, .N), "no") ] %>%
  .[]
#          date infection n_visits n_infect chronic
#        <Date>    <char>    <int>    <int>  <char>
# 1: 2005-01-01       yes        1        1    <NA>
# 2: 2005-06-30       yes        2        2    <NA>
# 3: 2005-10-15       yes        3        3     yes
# 4: 2006-01-01        no        4        3      no
# 5: 2006-06-01        no        4        2      no
# 6: 2006-11-01       yes        3        1      no
# 7: 2006-12-01        no        4        1      no
# 8: 2007-11-15       yes        2        1    <NA>

在回顾前几行的同时创建一个新变量

Creating a new variable while looking back at previous rows

datetime

r

lag

dplyr