在R中使用ifelse和grepl进行变异，并创建具有匹配字符串的新列

Question

我正在尝试使用grepl(部分匹配)创建新列，如果使用mutate，则创建新列。我有与单位的药物名称列表，我想创建新的列与这些单位。这是我的药名样本，

 Description
    ZOLEDRONIC ACID INF 4MG/5ML (5ML)
    CALCICHEW D3 FORTE CHEWABLE TABS
    DAYCASE DRUGS
    AKYNZEO (NETUPITANT/PALONOSETRON) HARD CAPS 300MG/0.5MG
    CYCLOPHOSPHAMIDE INF 1MG/ML
    EPIRUBICIN INJ 1MG/ML
    DAYCASE DRUGS
    DAYCASE DRUGS
    ALOXI (PALONOSETRON) INJ 250MCG
    PACLITAXEL INF 1MG/ML
    DEXAMETHASONE VIAL 6.6MG/2ML - USE PSD742
    PACLITAXEL INF 1MG/ML
    RAMUCIRUMAB INF 0.1MG - 100MG

这是我试过的密码

units <- c("TABS","INF","SYR","CAPS","INJ","VIAL","SOL","POWDER","GEL","CREAM","LOTION","AMP"," SINGLEJECT","PFS")

drug_name <- drug_name %>% mutate(unit =  grepl(paste(units,collapse = "|"),Description))

目前的输出是这样的，

Description                                 unit
ZOLEDRONIC ACID INF 4MG/5ML (5ML)           TRUE
    CALCICHEW D3 FORTE CHEWABLE TABS        TRUE
    DAYCASE DRUGS                           FALSE
    AKYNZEO (NETUPITANT/PALONOSETRON) CAPS  TRUE
    CYCLOPHOSPHAMIDE INF 1MG/ML             TRUE
    EPIRUBICIN INJ 1MG/ML                   TRUE
    DAYCASE DRUGS                           FALSE
    DAYCASE DRUGS                           FALSE

我不希望单位为真/假，相反，它应该指定确切的单位，如下，

        ZOLEDRONIC ACID INF 4MG/5ML (5ML)       INF
        CALCICHEW D3 FORTE CHEWABLE TABS        TABS

我正试着重新定义这个，但不知道该怎么做。另外，我需要将药物名称(4MG/5ML)中的指标提取到一个单独的列中。我可以在另一个查询中做到这一点，但需要使用单元进行单个查询。并不是所有的药物都有这种形式，有些是这样的-->01，of 100毫克，250微克。

            ZOLEDRONIC ACID INF 4MG/5ML (5ML)       INF    4MG/5ML
            CALCICHEW D3 FORTE CHEWABLE TABS        TABS   NA
            CYCLOPHOSPHAMIDE INF 1MG/ML             INF    1MG/ML
            ALOXI (PALONOSETRON) INJ 250MCG         INF    250MCG

有人能帮我解决这个问题吗？

Answer 1

潮汐溶液

pattern = paste( units, collapse = "|" )

df %>% mutate(medicine = stringr::str_split( Description, pattern, simplify = TRUE )[,1],
              medicine = stringr::str_trim( medicine ),
              unit =  stringr::str_extract( Description, pattern ),
              volume = stringr::str_split( Description, pattern, simplify = TRUE )[,2],
              volume = ifelse( volume == "", NA, volume ),
              volume = stringr::str_trim( volume ) ) %>%
  select( -Description )


#                                  medicine unit                 volume
# 1                         ZOLEDRONIC ACID  INF          4MG/5ML (5ML)
# 2             CALCICHEW D3 FORTE CHEWABLE TABS                   <NA>
# 3                           DAYCASE DRUGS <NA>                   <NA>
# 4  AKYNZEO (NETUPITANT/PALONOSETRON) HARD CAPS            300MG/0.5MG
# 5                        CYCLOPHOSPHAMIDE  INF                 1MG/ML
# 6                              EPIRUBICIN  INJ                 1MG/ML
# 7                           DAYCASE DRUGS <NA>                   <NA>
# 8                           DAYCASE DRUGS <NA>                   <NA>
# 9                    ALOXI (PALONOSETRON)  INJ                 250MCG
# 10                             PACLITAXEL  INF                 1MG/ML
# 11                          DEXAMETHASONE VIAL 6.6MG/2ML - USE PSD742
# 12                             PACLITAXEL  INF                 1MG/ML
# 13                            RAMUCIRUMAB  INF          0.1MG - 100MG

Answer 2

这里有一个使用dplyr和purrr的方法。

library(dplyr)
library(purrr)
drug_name <- data_frame(
  Description = trimws(readLines(textConnection('ZOLEDRONIC ACID INF 4MG/5ML (5ML)
    CALCICHEW D3 FORTE CHEWABLE TABS
    DAYCASE DRUGS
    AKYNZEO (NETUPITANT/PALONOSETRON) HARD CAPS 300MG/0.5MG
    CYCLOPHOSPHAMIDE INF 1MG/ML
    EPIRUBICIN INJ 1MG/ML
    DAYCASE DRUGS
    DAYCASE DRUGS
    ALOXI (PALONOSETRON) INJ 250MCG
    PACLITAXEL INF 1MG/ML
    DEXAMETHASONE VIAL 6.6MG/2ML - USE PSD742
    PACLITAXEL INF 1MG/ML
    RAMUCIRUMAB INF 0.1MG - 100MG')))
)
units <- c("TABS","INF","SYR","CAPS","INJ","VIAL","SOL","POWDER",
           "GEL","CREAM","LOTION","AMP","SINGLEJECT","PFS")
units1 <- sprintf("\\b(%s)\\b", paste(units, collapse = "|"))

首先，一个概念的穿透力和证据：

m <- regmatches(drug_name$Description,
                gregexpr(paste0(units1, ".*"), drug_name$Description))
m
# [[1]]
# [1] "INF 4MG/5ML (5ML)"
# [[2]]
# [1] "TABS"
# [[3]]
# character(0)
# [[4]]
# [1] "CAPS 300MG/0.5MG"
# [[5]]
# [1] "INF 1MG/ML"
# [[6]]
# [1] "INJ 1MG/ML"
# [[7]]
# character(0)
# [[8]]
# character(0)
# [[9]]
# [1] "INJ 250MCG"
# [[10]]
# [1] "INF 1MG/ML"
# [[11]]
# [1] "VIAL 6.6MG/2ML - USE PSD742"
# [[12]]
# [1] "INF 1MG/ML"
# [[13]]
# [1] "INF 0.1MG - 100MG"

为了找出单位的数量，我用.*“得到所有的跟随”。这有时会得到比我们想要的更多，但为了找到所有有价值的单词后的一个单位，我们需要抓住所有，然后过滤。

我们需要防范character(0)并在理想情况下将其转换为向量，因此

fixEmpties <- function(lst, default=NA_character_)
  unlist(replace(lst, lengths(lst) == 0L, default))

尽管您可能更喜欢使用空字符串""而不是NA_character_。

为了过滤，这里有一个快速的函数，它在下面的“单词”中查找任意数字。它还允许单独的"-"，但如果是最后一次匹配，则将其移除。

extractAmounts <- function(s, default=NA_character_) {
  vec <- strsplit(s, "\\s+")[[1]][-1]
  vec2 <- which(cumall(grepl("[0-9]", vec) | vec == "-"))
  if (isTRUE(length(vec2) > 0)) {
    # remove a trailing "-" that does not result in a range
    if (vec[ tail(vec2, 1) ] == "-") vec2 <- vec2[-length(vec2)]
    return(paste(vec[vec2], collapse = " "))
  } else return(default)
}

和一个快速得到原来的单位，总是第一个字。

extractUnits <- function(s)
  head(strsplit(s, "\\s+")[[1]], 1)

测试：

m <- fixEmpties(m)
sapply(m, extractUnits, USE.NAMES=FALSE)
#  [1] "INF"  "TABS" NA     "CAPS" "INF"  "INJ"  NA     NA     "INJ"  "INF" 
# [11] "VIAL" "INF"  "INF" 
sapply(m, extractAmounts, USE.NAMES=FALSE)
#  [1] "4MG/5ML (5ML)" NA              NA              "300MG/0.5MG"  
#  [5] "1MG/ML"        "1MG/ML"        NA              NA             
#  [9] "250MCG"        "1MG/ML"        "6.6MG/2ML"     "1MG/ML"       
# [13] "0.1MG - 100MG"

好吧，现在把这个放进管子里

drug_name %>%
  mutate(
    full = fixEmpties(regmatches(Description, gregexpr(paste0(units1, ".*"), Description))),
    unit = map_chr(full, extractUnits),
    amt = map_chr(full, extractAmounts)
  ) %>%
  select(-full)
# # A tibble: 13 x 3
#    Description                                             unit  amt          
#    <chr>                                                   <chr> <chr>        
#  1 ZOLEDRONIC ACID INF 4MG/5ML (5ML)                       INF   4MG/5ML (5ML)
#  2 CALCICHEW D3 FORTE CHEWABLE TABS                        TABS  <NA>         
#  3 DAYCASE DRUGS                                           <NA>  <NA>         
#  4 AKYNZEO (NETUPITANT/PALONOSETRON) HARD CAPS 300MG/0.5MG CAPS  300MG/0.5MG  
#  5 CYCLOPHOSPHAMIDE INF 1MG/ML                             INF   1MG/ML       
#  6 EPIRUBICIN INJ 1MG/ML                                   INJ   1MG/ML       
#  7 DAYCASE DRUGS                                           <NA>  <NA>         
#  8 DAYCASE DRUGS                                           <NA>  <NA>         
#  9 ALOXI (PALONOSETRON) INJ 250MCG                         INJ   250MCG       
# 10 PACLITAXEL INF 1MG/ML                                   INF   1MG/ML       
# 11 DEXAMETHASONE VIAL 6.6MG/2ML - USE PSD742               VIAL  6.6MG/2ML    
# 12 PACLITAXEL INF 1MG/ML                                   INF   1MG/ML       
# 13 RAMUCIRUMAB INF 0.1MG - 100MG                           INF   0.1MG - 100MG