R/import_excel_data.R

In agapiospanos/DDI: DDI criteria evaluation for potential interactions between drugs

#' Imports patient data from excel file.
#'
#' @param current_data (List) the data that was imported by another call of import_excel_data so that this call will append the new data to the old ones.
#' @param path (Character) the path that the excel file can be read from.
#' @param worksheet (Character) the worksheet that the desired variable is stored in.
#' @param var_col (Character) the column that stores the desired variable
#' @param ignore_na (Boolean) (optional) (default: FALSE) a boolean variable that indicates whether to ignore any N/As found in excel (if set to TRUE) or to handle them as missing data (if set to FALSE).
#' @param include_missing (Boolean) (optional) (default: FALSE) a boolean that indicated whether to include the missing values as NAs in the data to achieve one to one correspondance in the data if necessary.
#' @param excel_col_pid (Character) the column name that specifies the patient id in the excel you provided for the data.
#'
#' @return list of [[1]] patient data for the specified variable. [[2]] missing patient id and variable name as specified in excel that we used to import data.
#'
#' @author
#' Agapios Panos <panosagapios@gmail.com>
#'
#' @importFrom readxl read_excel

import_excel_data <- function(current_data, path, worksheet, var_col, ignore_na=FALSE, include_missing=FALSE, excel_col_pid = 'usubjid') {

    # specify the excel file location and worksheet
    sheet <- read_excel(path, worksheet)

    # stop execution and display error if the column name specified in var_col variable does not exist.
    if (!var_col %in% colnames(sheet)) {
        cat('An error occured while trying to parse data. Column ', var_col, ' not found at excel file you used to get data from. Please add data for this column or exclude this criterion.')
        stop('The specified column name ', var_col, ' is not present in the excel file that you tried to import... Please make sure you have correctly set the column names in the excel file')
    }

    if (!excel_col_pid %in% colnames(sheet)) {
        stop('The specified excel column ', excel_col_pid, ' for the patient id data is not present in the excel file. Please specify your own using the argument excel_col_pid or if you already did so, verify that it exists in the excel document you used to provide patient data.')
    }

    # ordering the sheet in ascending order by the specified column. Necessary for the proper parsing of the patient data as the algorithm expects the sheet data to be ordered and parses it in a serial way.
    sheet <- sheet[order(unlist(sheet[excel_col_pid])),]

    # storing the patients' ids
    pids <- unlist(sheet[excel_col_pid])

    # initializing the patient id for sheet using the first patient's id.
    pid <- pids[1]

    # initialize patient data
    pdata <- list()

    # initialize missing data for patients
    missing_data_patients <- list()

    # if the user specified current_data set them in pdata
    if (!missing(current_data)) {
        pdata <- current_data[[1]]
        missing_data_patients <- current_data[[2]]
    }

    # initializing the variable for the signle patient data.
    conditions <- c()

    # variable to check if this patient has missing data. Initialized to FALSE.
    has_missing <- FALSE

    for (i in 1:length(pids)) {
        if (pid == pids[i]) { # still the same patient id - collecting data

            #checking if it is NA - missing data in case we chose to not ignore them
            if (is.na(sheet[[var_col]][i])) {

                # if we will ignore the missing we will not store the data, neither we will set has_missing to TRUE
                if (!ignore_na) {
                    # update the logical var
                    has_missing <- TRUE
                }

                # if we must include the missing to achieve 1-1 correspondace we insert an NA
                if (include_missing) {
                    conditions <- c(conditions, NA)
                }

            } else { # is not NA data
                # storing the data to the list
                conditions <- c(conditions, sheet[[var_col]][i])
            }

        } else { # new patient id

            # keep the current data to the patient data var
            pdata[[pids[i - 1]]] <- c(pdata[[pids[i - 1]]], list(conditions)) # i-1 because we are refering to the previous patient's data

            # we also record the id if it has missing data.
            if (has_missing) {
                missing_data_patients[[pids[i - 1]]] <- c(missing_data_patients[[pids[i - 1]]], var_col) # i-1 because we are refering to the previous patient's data
            }

            # reseting the logical var for the next patient
            has_missing <- FALSE

            # assign the new id to the patient var so that we can compare to it in the next loop
            pid <- pids[i]

            # keep only the new data deleting the data from the old patient
            # checking if it is NA - missing data
            if (is.na(sheet[[var_col]][i])) {

                # if we will ignore the missing we will not store the data, neither we will set has_missing to TRUE and empty the list
                if (!ignore_na) {
                    # update the logical var
                    has_missing <- TRUE
                    # set conditions as an empty list
                    conditions <- c()
                }

                # if we must include the missing to achieve 1-1 correspondace we insert an NA
                if (include_missing) {
                    conditions <- c(NA)
                }

            } else {
                # storing the data to the list
                conditions <- c(sheet[[var_col]][i])
            }
        }
    }

    # keep the data for the last patient. Till now it kept the data for the n-1 patients
    pdata[[pids[i]]] <- c(pdata[[pids[i]]], list(conditions))

    # we also record the id if she/he has missing data.
    if (has_missing) {
        missing_data_patients[[pids[i]]] <- c(missing_data_patients[[pids[i]]], var_col)
    }

    # ret <- list(pdata, missing_data_patients)

    invisible(list(pdata, missing_data_patients))  # instead of return as we do not want to be printed
}