Create function to tidy CTgov data

There is now a bunch of code in functions/xml_parsing.R that makes tidy data of the baseline measurements in a clinical trial (without unit tests atm).

That code could easily be reworked into a function to do the same based on an XML. Would be cool.

# Initial xml parse
  x <- XML::xmlTreeParse(sel[i])
  # Get the root of the xml tree
  y <- XML::xmlRoot(x)
  z <- XML::xmlToList(y)

  nct_id <- z$id_info$nct_id
  study_design <- z$study_design
  randomized <- grepl(study_design, pattern = 'randomized', ignore.case = TRUE)
  start_date <- z$start_date
  completion_date <- z$completion_date
  overall_status <- z$overall_status
  trial_condition <- z$condition
  trial_phase <- z$phase
  oversight_authority <- z$oversight$authority
  trial_source <- z$source
  fda_regulated <- z$is_fda_regulated
  number_of_arms <- z$number_of_arms
  intervention_type <- z$intervention$intervention_type

  base_measures <- y[['clinical_results']][['baseline']][['measure_list']]
  df = data.frame(title = NULL, sub_title = NULL, units = NULL, param = NULL, group = NULL, value = NULL, spread = NULL)
  for (j in 1:length(base_measures))
  {
    base_measure <- base_measures[[j]]
    title <- base_measure[['title']][1]$text %>% as.character %>% tail(n = 1)
    units <- base_measure[['units']][1]$text %>% as.character %>% tail(n = 1)
    param <- base_measure[['param']][1]$text %>% as.character %>% tail(n = 1)
    cat_count <- length(base_measure[['category_list']])
    if (cat_count > 1)
    {
      for (cat in 1:cat_count)
      {
        base_me <- base_measure[['category_list']][cat]
        sub_title <- base_me$category[['sub_title']][1]$text %>% as.character %>% tail(n = 1)
        group_count <- length(base_me$category[['measurement_list']])
        for (q in 1:group_count)
        {
           tmp <- XML::xmlToList(base_me$category[['measurement_list']][q]$measurement)
           group <- tmp['group_id'] %>% as.character
           value <- tmp['value'] %>% as.numeric
           spread <- tmp['spread'] %>% as.numeric
           print(c(title, units, param, group, value, spread))
           df <- rbind(df, data.frame(title, sub_title, units, param, group, value, spread))
        }
      }

    } else
    {
      group_count <- length(base_measure[['category_list']][['category']][['measurement_list']])
      for (q in 1:group_count)
      {
         tmp <- XML::xmlToList(base_measure[['category_list']][['category']][['measurement_list']][q]$measurement)
         group <- tmp['group_id'] %>% as.character
         value <- tmp['value'] %>% as.numeric
         spread <- tmp['spread'] %>% as.numeric
         print(c(title, units, param, group, value, spread))
         df <- rbind(df, data.frame(title, sub_title = NA, units, param, group, value, spread))
      }
    }

  }

Just putting it in issue as a to-do and so that anyone who feels like contributing sees this as an opportunity :-)

chartgerink / 2017ctgov

Create function to tidy CTgov data #5