Create new variable based on outcome of other variable in group - R

This a similar/followup question to this <https://stackoverflow.com/questions/75475934/r-how-to-code-new-variable-based-on-grouped-variable-and-conditioned-on-earlier> but it is different because within donors there are potentially two match runs.

I have a data file with organ donors. I'm looking at lungs that are donated - there are two lungs.

If the lungs are split (L and R) and put up for donation, they are each attempted to match with recipients ("matchrun"). They go through eligible recipients until one matches ("sequence").

If the lung is matched to a recipient, it goes to them ("organ_placed").

If the lung doesn't match, it continues in the sequence and then just remains NA at the maximum sequence number.

I would like to create a new variable that has the outcome of the match run such that if one lung is placed and the other is not, it tells you that the lung was discarded. i.e. see case of Donor 2 in the data - the left lung is placed, but the right doesn't match.

In donor 3, the first match run doesn't match but the match run for the other lung does.

I figure it will be something like group_by(donorid, matchrun) but then how do you make a condition based on the match run?

library(tribble)
library(dplyr)

data <- tribble(
  ~donorid, ~matchrun, ~sequence, ~organ_placed,
    2, 3, 1, NA,
  2, 3, 2, NA,
  2, 3, 3, "L",
  2, 4, 1, NA,
  2, 4, 2, NA,
  2, 4, 3, NA,
  3, 5, 1, NA,
  3, 5, 1, NA,
  3, 5, 1, NA,
  3, 6, 1, NA,
  3, 6, 2, NA,
  3, 6, 3, "L"
)

desired_outcome <- tribble(
  ~donorid, ~matchrun, ~sequence, ~organ_placed, ~organ,
  2, 3, 1, NA, NA, 
  2, 3, 2, NA, NA, 
  2, 3, 3, "L", "Left Single",
  2, 4, 1, NA, NA,
  2, 4, 2, NA, NA, 
  2, 4, 3, NA, "Right Discarded",
  3, 5, 1, NA, NA,
  3, 5, 1, NA, NA,
  3, 5, 1, NA, "Right Discarded",
  3, 6, 1, NA, NA,
  3, 6, 2, NA, NA,
  3, 6, 3, "L", "Left Single")

You can try this:

data %>% 
  group_by(donorid) %>% 
  mutate(temp = ifelse(n_distinct(organ_placed, na.rm = TRUE) == 1, unique(na.omit(organ_placed)), "B")) %>% 
  group_by(matchrun, .add = TRUE) %>% 
  mutate(organ = case_when(organ_placed == "L" ~ "Left Single",
                           organ_placed == "R" ~ "Right Single",
                           all(is.na(organ_placed)) & row_number() == max(sequence) & temp == "L" ~ "Right Discarded", 
                           all(is.na(organ_placed)) & row_number() == max(sequence) & temp == "R" ~ "Left Discarded")) %>%
  ungroup()

output

   donorid matchrun sequence organ_placed temp  organ       
 1       1        1        1 NA           B     NA          
 2       1        1        2 NA           B     NA          
 3       1        1        3 L            B     Left Single 
 4       1        2        1 NA           B     NA          
 5       1        2        2 NA           B     NA          
 6       1        2        3 R            B     Right Single
 7       2        3        1 NA           L     NA          
 8       2        3        2 NA           L     NA          
 9       2        3        3 L            L     Left Single 
10       2        4        1 NA           L     NA          
11       2        4        2 NA           L     NA          
12       2        4        3 NA           L     Right Discarded

We can use

library(data.table)
library(stringr)
setDT(data)[, seq2 := rowid(donorid, matchrun) ]
data[, organ := str_replace_all(organ_placed,
   setNames(c("Left Single", "Right Single"), c("L", "R")))]
 data[seq2 == max(seq2), 
  organ := fcase(!is.na(organ), organ, default = 
  str_replace_all(setdiff(c("Left Single", "Right Single"), organ), 
   setNames(c("Left Discarded", "Right Discarded"),
   c("Left Single", "Right Single")))), donorid
  ][, seq2 := NULL][]

-output

> data
    donorid matchrun sequence organ_placed           organ
 1:       2        3        1         <NA>            <NA>
 2:       2        3        2         <NA>            <NA>
 3:       2        3        3            L     Left Single
 4:       2        4        1         <NA>            <NA>
 5:       2        4        2         <NA>            <NA>
 6:       2        4        3         <NA> Right Discarded
 7:       3        5        1         <NA>            <NA>
 8:       3        5        1         <NA>            <NA>
 9:       3        5        1         <NA> Right Discarded
10:       3        6        1         <NA>            <NA>
11:       3        6        2         <NA>            <NA>
12:       3        6        3            L     Left Single

Update: we have to add matchrun to the group. Removed prior solution:

data %>% 
  group_by(donorid, matchrun) %>% 
  mutate(outcome = case_when(organ_placed == "L" ~ "Left Single",
                             organ_placed == "R" ~ "Right Single",
                             organ_placed == "B" ~ "Bilateral",
                             (is.na(organ_placed) & 
                                row_number() == max(row_number())) & 
                               "L" %in% organ_placed ~ "Right Discarded",
                             (is.na(organ_placed) & 
                                row_number() == max(row_number())) & 
                               "R" %in% organ_placed ~ "Left Discarded",
                             TRUE ~ NA_character_))

Groups:   donorid, matchrun [4]
   donorid matchrun sequence organ_placed outcome    
     <dbl>    <dbl>    <dbl> <chr>        <chr>      
 1       2        3        1 NA           NA         
 2       2        3        2 NA           NA         
 3       2        3        3 L            Left Single
 4       2        4        1 NA           NA         
 5       2        4        2 NA           NA         
 6       2        4        3 NA           NA         
 7       3        5        1 NA           NA         
 8       3        5        1 NA           NA         
 9       3        5        1 NA           NA         
10       3        6        1 NA           NA         
11       3        6        2 NA           NA         
12       3        6        3 L            Left Single