• Disaggregation of ASA24 Foods
• SCRIPT
  • Dietary Data Filtering
  • Specify grouping variables
  • Sum by Entry (Recall/Record) for total kcal and other nutrient totals
  • Minimize the number of columns to the essential data
  • Apply Ingredient Percentage Adjustment for Coffee and Tea Brewing
  • Disaggregate Food Codes and compute final Ingredient Weights
  • Write output files

Disaggregation of ASA24 Foods

This script takes in ASA24 ITEMS files, disaggregates WWEIA food codes to ingredients, and calculates the new ingredient weight. This script also calculates total caloric intake & other nutrients for each participant recall so polyphenol intakes can be standardized to caloric intake later on.

INPUTS

• Your Dietary Data - This script does not provide filtering for portion or nutrient outliers. These may be performed in advance. The ASA24 website has cleaning recommendations here: “Reviewing and Cleaning ASA24 Data”.

• FDA_FDD_All_Records_v_3.1.xlsx - FDD FoodCodes to Ingredients and Ingredient Percentages

OUTPUTS

• Diet_Disaggregated.csv.bz2: Dietary data that has been disaggregated using FDD.
• Diet_total_nutrients.csv: Total daily kcal intakes for unique records (subject, RecallNo for Recall; subject, RecordNo, RecordDayNo for Record)

SCRIPT

# Load packages
suppressMessages(library(dplyr))
suppressMessages(library(vroom))
suppressMessages(library(tidyr))
suppressMessages(library(stringr))
suppressMessages(library(readxl))

Load Example Dietary Data and FDA-FDD V3.6

# Load provided file paths
source("provided_files.R")

# Load User Dietary Data
input_data = vroom::vroom(diet_input_file, show_col_types = FALSE) %>%
  dplyr::rename(subject = UserName)

# FDD Disaggregation options
# Rename for ease of use.
FDD_V3 = readxl::read_xlsx(FDD_file) %>%
  dplyr::rename(latest_survey = "Latest Survey",
         wweia_food_code = "WWEIA Food Code",
         wweia_food_description = "WWEIA Food Description",
         fdd_ingredient = "Basic Ingredient Description",
         ingredient_percent = "Ingredient Percent") %>%
  dplyr::select(wweia_food_code, wweia_food_description,  
         fdd_ingredient, ingredient_percent) %>%
  dplyr::mutate(wweia_food_code = as.integer(wweia_food_code))

Dietary Data Filtering

Filter IN Individuals With More than Recall/Record and filter OUT incomplete recalls/Records

• RecallStatus: 2=Complete; 5=Breakoff/Quit
• RecordDayStatus: 2=Complete; 5=Breakoff/Quit

if ("RecallNo" %in% names(input_data)) {
  
  # Recall dataset
  input_data_clean = input_data %>%
    group_by(subject) %>%
    filter(n_distinct(RecallNo) > 1) %>%
    ungroup() %>%
    filter(RecallStatus != 5)
  
} else if ("RecordNo" %in% names(input_data)) {
  
  # Record dataset
  input_data_clean = input_data %>%
    group_by(subject) %>%
    filter(n_distinct(RecordNo) > 1 | n_distinct(RecordDayNo) > 1) %>%
    ungroup() %>%
    filter(RecordDayStatus != 5)
  
} else {
  stop("Data must contain RecallNo or RecordNo.")
}

Specify grouping variables

Column grouping depends on whether ASA24 output is from a record or recall.

if ("RecallNo" %in% names(input_data_clean)) {
  group_vars = c("subject", "RecallNo")
  
} else if ("RecordNo" %in% names(input_data_clean)) {
  group_vars = c("subject", "RecordNo", "RecordDayNo")
  
} else {
  stop("Data must contain RecallNo or RecordNo.")
}

Sum by Entry (Recall/Record) for total kcal and other nutrient totals

input_total_nutrients = input_data_clean %>%
  dplyr::group_by(across(all_of(group_vars))) %>%
  dplyr::summarize(across(KCAL:B12_ADD, ~ sum(.x, na.rm = TRUE), 
                          .names = "Total_{.col}")) %>%
  dplyr::ungroup()

## `summarise()` has grouped output by 'subject'. You can override using the
## `.groups` argument.

Minimize the number of columns to the essential data

input_data_clean_minimal = input_data_clean %>%
  dplyr::rename(wweia_food_code = FoodCode,
         food_description = Food_Description) %>%
  dplyr::select(c(subject,
                  any_of(c("RecallNo", "RecordNo", "RecordDayNo")),
                  wweia_food_code, food_description, FoodAmt))

Apply Ingredient Percentage Adjustment for Coffee and Tea Brewing

FDD_V3_adjusted = FDD_V3 %>%
  dplyr::group_by(wweia_food_code) %>%
  dplyr::mutate(
    
    # Create Flag 
    has_tea    = any(str_detect(fdd_ingredient, regex("Tea", ignore_case = TRUE))),
    has_coffee = any(str_detect(fdd_ingredient, regex("Coffee", ignore_case = TRUE))),
    has_water  = any(str_detect(fdd_ingredient, regex("Water", ignore_case = TRUE))),

    # Add combined coffee|tea + water percentages
    brewing_adjustment_total = case_when(
      # Tea + Water
      has_tea & has_water ~ sum(
        ingredient_percent[str_detect(fdd_ingredient, regex("Tea|Water", ignore_case = TRUE))],
        na.rm = TRUE),
      # Coffee + Water
      has_coffee & has_water ~ sum(
        ingredient_percent[str_detect(fdd_ingredient, regex("Coffee|Water", ignore_case = TRUE))],
        na.rm = TRUE),
      TRUE ~ NA_real_),
    
    # Ensure new brewing adjustment percentage is applied to only coffee|tea
    brewing_adjustment_percentage = if_else(
      str_detect(fdd_ingredient, regex("Coffee|Tea", ignore_case = TRUE)),
      brewing_adjustment_total,
      NA_real_)) %>%
  dplyr::select(-c(has_tea, has_coffee, has_water, brewing_adjustment_total)) %>%
  dplyr::ungroup()

Disaggregate Food Codes and compute final Ingredient Weights

merge = dplyr::left_join(input_data_clean_minimal, FDD_V3_adjusted, by = "wweia_food_code", 
                  relationship = "many-to-many") %>%
  # Compute final ingredient weight
  # If brewing adjustment exists, it will use this adjustment first.
  dplyr::mutate(FoodAmt_Ing_g = FoodAmt * (
      coalesce(brewing_adjustment_percentage, ingredient_percent) / 100))

Write output files

Ensure outputs directory is created

if (!dir.exists("outputs")) dir.create("outputs", recursive = TRUE)

Write Files

vroom::vroom_write(merge, 'outputs/Diet_Disaggregated.csv.bz2', delim = ",")
vroom::vroom_write(input_total_nutrients, 'outputs/Diet_total_nutrients.csv', delim = ",")