嗨，我需要用R编程语言解决一个问题
library(gtools)
end_date <- "2021-12-31"
ddf1 <- data.frame(pnr=c("1","1","2","2","3","3","3","4"),
in_date=as.POSIXct(c("2010-08-18","2010-09-01","2019-04-02","2018-03-27",
"2019-07-12","2013-10-20","2012-07-01","2015-05-02")),
out_date=as.POSIXct(c("2010-12-04",NA,"2019-05-17",NA,
NA,"2017-08-19",NA,NA)),
treat1=c(1,1,1,1,1,1,1,1)
)
ddf2 <- data.frame(pnr=c("4","4","3","3","2","2","2","1"),
in_date=as.POSIXct(c("2010-08-18","2010-09-01","2019-04-02","2018-03-27",
"2019-07-12","2013-10-20","2012-07-01","2015-05-02")),
out_date=as.POSIXct(c("2010-12-04",NA,"2019-05-17",NA,
NA,"2017-08-19",NA,NA)),
treat2=c(1,1,1,1,1,1,1,1)
)
expected_output1 <- data.frame(pnr=c("1","2","3","3","4"),
in_date=as.POSIXct(c("2010-08-18","2018-03-27",
"2019-07-12","2012-07-01","2015-05-02")),
out_date=as.POSIXct(c("2010-12-04","2019-05-17",end_date,
"2017-08-19",end_date)),
treat1=c(1,1,1,1,1)
)
expected_output2 <- data.frame(pnr=c("4","3","2","2","1"),
in_date=as.POSIXct(c("2010-08-18","2018-03-27",
"2019-07-12","2012-07-01","2015-05-02")),
out_date=as.POSIXct(c("2010-12-04","2019-05-17",end_date,
"2017-08-19",end_date)),
treat2=c(1,1,1,1,1)
)
ddf <- smartbind(ddf1,ddf2)
expected_output <- smartbind(expected_output1,expected_output2)
> ddf
pnr in_date out_date treat1 treat2
1:1 1 2010-08-18 2010-12-04 1 NA
1:2 1 2010-09-01 <NA> 1 NA
1:3 2 2019-04-02 2019-05-17 1 NA
1:4 2 2018-03-27 <NA> 1 NA
1:5 3 2019-07-12 <NA> 1 NA
1:6 3 2013-10-20 2017-08-19 1 NA
1:7 3 2012-07-01 <NA> 1 NA
1:8 4 2015-05-02 <NA> 1 NA
2:1 4 2010-08-18 2010-12-04 NA 1
2:2 4 2010-09-01 <NA> NA 1
2:3 3 2019-04-02 2019-05-17 NA 1
2:4 3 2018-03-27 <NA> NA 1
2:5 2 2019-07-12 <NA> NA 1
2:6 2 2013-10-20 2017-08-19 NA 1
2:7 2 2012-07-01 <NA> NA 1
2:8 1 2015-05-02 <NA> NA 1
> expected_output
pnr in_date out_date treat1 treat2
1:1 1 2010-08-18 2010-12-04 1 NA
1:2 2 2018-03-27 2019-05-17 1 NA
1:3 3 2019-07-12 2021-12-31 1 NA
1:4 3 2012-07-01 2017-08-19 1 NA
1:5 4 2015-05-02 2021-12-31 1 NA
2:1 4 2010-08-18 2010-12-04 NA 1
2:2 3 2018-03-27 2019-05-17 NA 1
2:3 2 2019-07-12 2021-12-31 NA 1
2:4 2 2012-07-01 2017-08-19 NA 1
2:5 1 2015-05-02 2021-12-31 NA 1
我有一些人经历了不同的治疗，treat1和treat2。我需要处理这样一个事实，即一些治疗课程已经开始，但缺少out_date。在缺少out_date的情况下，该研究应该用end_date替换：
end_date <- "2021-12-31"
然而，如果观察结果与
pnr in_date out_date treat1 treat2 1:1 1 2010-08-18 2010-12-04 1 NA 1:2 1 2010-09-01 <NA> 1 NA
如果in_date（表示治疗的开始）在同一个人的另一次治疗期间内或“pnr”，则正确的输出为：
pnr in_date out_date treat1 treat2 1:1 1 2010-08-18 2010-12-04 1 NA
因为2010-08-18是最早的in_date。然而，如果行中有更早的日期而没有out_date，则应使用该日期，这是pnr 2的情况
pnr in_date out_date treat1 treat2 1:3 2 2019-04-02 2019-05-17 1 NA 1:4 2 2018-03-27 <NA> 1 NA
变为：
1:2 2 2018-03-27 2019-05-17 1 NA
因此涵盖了整个治疗期。
在没有out_date的情况下，应改为设置end_date；以便：
pnr in_date out_date treat1 treat2 1:8 4 2015-05-02 <NA> 1 NA
变为：
1:5 4 2015-05-02 2021-12-31 1 NA
在较早的日期或相交的日期以及缺少out_date的较晚的in_date的特殊情况下，函数应该能够处理它，如pnr 3
1:5 3 2019-07-12 <NA> 1 NA 1:6 3 2013-10-20 2017-08-19 1 NA 1:7 3 2012-07-01 <NA> 1 NA
应变成：
pnr in_date out_date treat1 treat2 1:3 3 2019-07-12 2021-12-31 1 NA 1:4 3 2012-07-01 2017-08-19 1 NA
可选：如果这是可能的，那么如果函数可以根据不同的处理方式来不同地处理这一点，那就太好了，因此每个pnr在每个treat1和treat2中的处理方式也不同，如expected_out中所示
我试图编写一些代码来比较out_date是否为NA，以及日期之间的差异，但我仍然无法理解如何继续：
ddf$end_replaced <- as.integer(ifelse(is.na(ddf$out_date),1,0))
ddf <- data.table(ddf)
ddf <- ddf[order(ddf$treat1,ddf$pnr,ddf$in_date,ddf$out_date),]
ddf[, diffx := difftime(in_date, shift(in_date, fill=in_date[1L]),
units="days"), by=pnr]
感谢阅读