How to filter alphanumeric characters range?

Question

I need to create dummy variables using ICD-10 codes. For example, chapter 2 starts with C00 and ends with D48X. Data looks like this:

data <- data.frame(LINHAA1 = c("B342", "C000", "D450", "0985"),
                   LINHAA2 = c("U071", "C99", "D68X", "J061"),
                   LINHAA3 = c("D48X", "Y098", "X223", "D640"))

Then I need to create a column that receives 1 if it's between the C00-D48X range and 0 if it's not. The result I desire:

LINHAA1   LINHAA2   LINHAA3  CHAPTER2
B342      U071      D48X         1
C000      C99       Y098         1
D450      D68X      X223         1
O985      J061      D640         0

It needs to go through LINHAA1 to LINHAA3. Thanks in advance!

Are you sure the data.frame should look like this? You called the variables "LinhaX", which is portuguese for "LineX" yet, as the toy data.frame was created, those are actually columns/variables — GuedesBF, Jan 29 '23 at 19:27
To create The column CHAPTER2, which other variable/lines are you checking against the target interval? — GuedesBF, Jan 29 '23 at 19:29
@GuedesBF "Linha" is the name of the column/variable because I'm working with death certificates, and they have Linha A, Linha B, Linha C and Linha D. When it becomes a database, they turn into columns. — Lana Meijinhos, Jan 29 '23 at 19:31
@GuedesBF to create CHAPTER2 I need to check all LINHA columns from each row. — Lana Meijinhos, Jan 29 '23 at 19:33
Do you mean if _any one of_ LINHAA1, LINHAA2, or LINHAA3 is in the range the column gets a 1? — G5W, Jan 29 '23 at 19:33
Ok, @LanaMeijinhos, acho que entendi, veja se a resposta ajuda — GuedesBF, Jan 29 '23 at 20:05

score 3 · Accepted Answer · answered Jan 29 '23 at 19:39

This should do it:

as.numeric(apply(apply(data, 1, 
    function(x) { x >="C00" & x <= "D48X" }), 2, any))
[1] 1 1 1 0

A little explanation: Checking if the codes are in the range can just be checked using alphabetic order (which you can get from <= etc). The inner apply checks each element and produces a matrix of logical values. The outer apply uses any to check if any one of the three logical values is true. as.numeric changes the result from TRUE/False to 1/0.

GuedesBF · Answer 2 · 2023-01-29T20:33:53.070

This is the typical case for dplyr::if_any. if_any returns TRUE if a given condition is met in any of the tested columns, rowwise:

library(dplyr)

data %>%
    mutate(CHAPTER2 = +if_any(starts_with("LINHAA"),
                             ~.x >= 'C00' & .x <='D48X'))

  LINHAA1 LINHAA2 LINHAA3 CHAPTER2
1    B342    U071    D48X        1
2    C000     C99    Y098        1
3    D450    D68X    X223        1
4    0985    J061    D640        0

zx8754 · Answer 3 · 2023-01-29T20:51:51.600

Using dedicated icd package

# remotes::install_github("jackwasey/icd")
library(icd)

#get the 2nd chapter start and end codes
ch2 <- icd::icd10_chapters[[ 2 ]]
# start   end 
# "C00" "D49" 

#expland the codes to include all chapter2 codes
ch2codes <- expand_range(ch2[ "start" ], ch2[ "end" ])
# length(ch2codes)
# 2094

#check if codes in a row match
ix <- apply(data, 1, function(i) any(i %in% ch2codes))
# [1] FALSE  TRUE FALSE FALSE

data$chapter2 <- as.integer(ix)
#data
#   LINHAA1 LINHAA2 LINHAA3 chapter2
# 1    B342    U071    D48X        0
# 2    C000     C99    Y098        1
# 3    D450    D68X    X223        0
# 4    0985    J061    D640        0

Note that you have some invalid codes:

#invalid
is_defined("D48X")
# [1] FALSE
explain_code("D48X")
# character(0)

#Valid
is_defined("D48")
# [1] TRUE
explain_code("D48")
# [1] "Neoplasm of uncertain behavior of other and unspecified sites"

How to filter alphanumeric characters range?

3 Answers3