Skip to content

Latest commit

 

History

History
243 lines (194 loc) · 7.08 KB

README.md

File metadata and controls

243 lines (194 loc) · 7.08 KB
Raw

MEDDRA API with R

Collection of functions to work with the MEDDRA.org api to code and classify adverse events in clinical trials.

The MEDDRA API is well documented in their developer documentation available once an account is created and on their public (swagger website)[https://mapisbx.meddra.org/index.html]. To use the swagger website you do need to have an account.

While working with this API, I was unable to find any example code to work with the API in R so publishing this in order to hopefully save others the headache.

What these functions offer is a mechanism to construct a valid json submission to the api. Other functions required that a json file be constructed externally and sent.

Dependancies

The functions used here make use of functions from:

  • httr
  • jsonlite
  • dplyr

The dmtools package for R was initially used to interact with the MEDDRA API but these functions did not sufficiently address the submission requirements for the current MEDDRA api which requies a json query structure. The authentication functions in this repo reproduce the underlying authentication methods from dmtools.

Functions:

API status

This function does not require authentication.

meddra_api_status <- function(){
  api_status <- content(GET(url = "https://mapisbx.meddra.org/api/status",
                            add_headers(accept = "*/*")),
                        "text")

  print(api_status, quote = FALSE )
}

Identify the ontology level of an unknown MEDDRA code e.g. PT, LLT

This does not require authentication

check_meddra_term_level <- function(medra_code) {
  term_level <- suppressMessages(
    httr::content(httr::GET(
      url = paste(
        "https://mapisbx.meddra.org/api/hist",
        medra_code,
        1,
        "English",
        "Release",
        sep = "/"
      ),
      httr::add_headers(
        "accept" = "application/json",
        "Content-Type" = "text/json",
        "Authorization" = paste("Bearer", token)
      )
    ),
    "text") %>%
      jsonlite::fromJSON() %>%
      # obtaining the term level from the latest version of MEDDRA
      dplyr::arrange(desc(medDraVer)) %>%
      dplyr::slice(1) %>%
      dplyr::select(termLevel)
  )
  return(term_level$termLevel)
}

Obtain MEDRA authentication token

This function requires that the used has a registration with the MEDDRA service and has obtained an API key

meddra_obtain_token <- function(){

  url <- 'https://mid.meddra.org/connect/token'

  meddra_id <- rstudioapi::askForPassword(prompt = "Please provide MEDRA USER ID")

  api_key <- rstudioapi::askForPassword(prompt = "Please provide MEDRA API Key")

  # token <- meddra_auth(token_url, meddra_id, api_key)

  # The below code is adapted from the authentication functions in dmtools
  token <- httr::POST(
    url = url,
    body = list(grant_type = "password", username = meddra_id, password = api_key, scope = "meddraapi"),
    httr::authenticate("mspclient", "clientsecret"),
    encode = "form"
  )
token <- httr::content(token)$access_token
  return(token)
}

token <- meddra_obtain_token()

Obtain term level of MEDDRA code

This function requires that the token has been obtained using the meddra_obtain_token function above and assigned to token

check_meddra_term_level <- function(medra_code) {
  term_level <- suppressMessages(
    httr::content(httr::GET(
      url = paste(
        "https://mapisbx.meddra.org/api/hist",
        medra_code,
        1,
        "English",
        "Release",
        sep = "/"
      ),
      httr::add_headers(
        "accept" = "application/json",
        "Content-Type" = "text/json",
        "Authorization" = paste("Bearer", token)
      )
    ),
    "text") %>%
      jsonlite::fromJSON() %>%
      # obtaining the term level from the latest version of MEDDRA
      dplyr::arrange(desc(medDraVer)) %>%
      dplyr::slice(1) %>%
      dplyr::select(termLevel)
  )
  return(term_level$termLevel)
}

Obtain term label from MEDDRA API

This function needs to have a logical evaluation of the term level implemented using the check_meddra_term_level function above to correctly create the json model submission

meddra_obtain_label <- function(medra_code){

 json_output <- list(
      bview = jsonlite::unbox("SOC"),
      rsview = jsonlite::unbox("Release"),
      code = jsonlite::unbox(0),
      pcode = jsonlite::unbox(0),
      syncode = jsonlite::unbox(0),
      lltcode = jsonlite::unbox(0),
      ptcode = jsonlite::unbox(medra_code),
      hltcode = jsonlite::unbox(0),
      hlgtcode = jsonlite::unbox(0),
      soccode = jsonlite::unbox(0),
      smqcode = jsonlite::unbox(0),
      type = jsonlite::unbox("PT"),
      addlangs = list(),
      rtype = jsonlite::unbox("M"),
      lang = jsonlite::unbox("English"),
      ver = jsonlite::unbox(20.1),
      kana = jsonlite::unbox(FALSE),
      separator = jsonlite::unbox(0)
    ) %>% jsonlite::toJSON()

   url <- "https://mapisbx.meddra.org/api/type"

  # Submit the call to the MEDDRA API
  response <- httr::POST(
    url,
    httr::add_headers(
      "accept" = "application/json",
      "Content-Type" = "text/json",
      "Authorization" = paste("Bearer", token)
    ),
    body = json_output
  )
# Take the returned json convert to df
  search_result <- httr::content(response, "text") %>%
    jsonlite::fromJSON() %>%
    as.data.frame()%>%
    # Some pt terms have multiple SOC this returns the primary allocation
    dplyr::filter(primarysocfg == "Y")
  # obtain the ptname for the medra code
  label <- search_result$ptname

  return(label)

}

Obtain term SOC from MEDDRA API

This function repeats the label methods but extracts and returns the SOC this could be encorporated into a function argument

meddra_obtain_soc <- function(medra_code){

  json_output <- list(
    bview = jsonlite::unbox("SOC"),
    rsview = jsonlite::unbox("Release"),
    code = jsonlite::unbox(0),
    pcode = jsonlite::unbox(0),
    syncode = jsonlite::unbox(0),
    lltcode = jsonlite::unbox(0),
    ptcode = jsonlite::unbox(medra_code),
    hltcode = jsonlite::unbox(0),
    hlgtcode = jsonlite::unbox(0),
    soccode = jsonlite::unbox(0),
    smqcode = jsonlite::unbox(0),
    type = jsonlite::unbox("PT"),
    addlangs = list(),
    rtype = jsonlite::unbox("M"),
    lang = jsonlite::unbox("English"),
    ver = jsonlite::unbox(20.1),
    kana = jsonlite::unbox(FALSE),
    separator = jsonlite::unbox(0)
  ) %>% jsonlite::toJSON()

  url <- "https://mapisbx.meddra.org/api/type"

  # Submit the call to the MEDDRA API
  response <- httr::POST(
    url,
    httr::add_headers(
      "accept" = "application/json",
      "Content-Type" = "text/json",
      "Authorization" = paste("Bearer", token)
    ),
    body = json_output
  )
  # Take the returned json convert to df
  search_result <- httr::content(response, "text") %>%
    jsonlite::fromJSON() %>%
    as.data.frame() %>%
    # Some pt terms have multiple SOC this returns the primary allocation
    dplyr::filter(primarysocfg == "Y")
  # obtain the soc name for the meddra code
  label <- search_result$socname

  return(label)

}